feat(ai): add code interpreter for AI data analysis (#16559)

## Summary

- Add code interpreter tool that enables AI to execute Python code for
data analysis, CSV processing, and chart generation
- Support for both local (development) and E2B (sandboxed production)
execution drivers
- Real-time streaming of stdout/stderr and generated files
- Frontend components for displaying code execution results with
expandable sections

## Code Quality Improvements

- Extract `getMimeType` to shared utility to reduce code duplication
between drivers
- Fix security issue: escape single quotes/backslashes in E2B driver env
variable injection
- Add `buildExecutionState` helper to reduce duplicated state object
construction
- Add `DEFAULT_CODE_INTERPRETER_TIMEOUT_MS` constant for consistency
- Fix lingui linting warning and TypeScript theme errors in frontend

## Test Plan

- [ ] Test code interpreter with local driver in development
- [ ] Test code interpreter with E2B driver in production environment
- [ ] Verify streaming output displays correctly in chat UI
- [ ] Verify generated files (charts, CSVs) are uploaded and
downloadable
- [ ] Test file upload flow (CSV, Excel) triggers code interpreter

<!-- CURSOR_SUMMARY -->
---

> [!NOTE]
> Updates generated i18n catalogs for Polish and pseudo-English, adding
strings for code execution/output (code interpreter) and various UI
messages, with minor text adjustments.
> 
> - **Localization**:
> - **Generated catalogs**: Refresh `locales/generated/pl-PL.ts` and
`locales/generated/pseudo-en.ts`.
> - Add strings for code execution/output (e.g., code, copy code/output,
running/waiting states, download files, generated files, Python code
execution).
> - Include new UI texts (errors, prompts, menus) and minor text
corrections.
>   - No changes to `pt-BR`; other files unchanged functionally.
> 
> <sup>Written by [Cursor
Bugbot](https://cursor.com/dashboard?tab=bugbot) for commit
befc13d02c. This will update automatically
on new commits. Configure
[here](https://cursor.com/dashboard?tab=bugbot).</sup>
<!-- /CURSOR_SUMMARY -->

packages/twenty-server/src/engine/core-modules/code-interpreter/code-interpreter-module.factory.ts

@@ -0,0 +1,53 @@
+import { NodeEnvironment } from 'src/engine/core-modules/twenty-config/interfaces/node-environment.interface';
+
+import { type TwentyConfigService } from 'src/engine/core-modules/twenty-config/twenty-config.service';
+
+import {
+  CodeInterpreterDriverType,
+  type CodeInterpreterModuleOptions,
+} from './code-interpreter.interface';
+
+export const codeInterpreterModuleFactory = async (
+  twentyConfigService: TwentyConfigService,
+): Promise<CodeInterpreterModuleOptions> => {
+  const driverType = twentyConfigService.get('CODE_INTERPRETER_TYPE');
+  const timeoutMs = twentyConfigService.get('CODE_INTERPRETER_TIMEOUT_MS');
+
+  switch (driverType) {
+    case CodeInterpreterDriverType.LOCAL: {
+      const nodeEnv = twentyConfigService.get('NODE_ENV');
+
+      if (nodeEnv === NodeEnvironment.PRODUCTION) {
+        throw new Error(
+          'LOCAL code interpreter driver is not allowed in production. Use E2B driver instead by setting CODE_INTERPRETER_TYPE=E2B and providing E2B_API_KEY.',
+        );
+      }
+
+      return {
+        type: CodeInterpreterDriverType.LOCAL,
+        options: { timeoutMs },
+      };
+    }
+    case CodeInterpreterDriverType.E2B: {
+      const apiKey = twentyConfigService.get('E2B_API_KEY');
+
+      if (!apiKey) {
+        throw new Error(
+          'E2B_API_KEY is required when CODE_INTERPRETER_TYPE is E2B',
+        );
+      }
+
+      return {
+        type: CodeInterpreterDriverType.E2B,
+        options: {
+          apiKey,
+          timeoutMs,
+        },
+      };
+    }
+    default:
+      throw new Error(
+        `Invalid code interpreter driver type (${driverType}), check your .env file`,
+      );
+  }
+};

packages/twenty-server/src/engine/core-modules/code-interpreter/code-interpreter.constants.ts

@@ -0,0 +1,3 @@
+export const CODE_INTERPRETER_DRIVER = Symbol('CODE_INTERPRETER_DRIVER');
+
+export const DEFAULT_CODE_INTERPRETER_TIMEOUT_MS = 300_000;

packages/twenty-server/src/engine/core-modules/code-interpreter/code-interpreter.interface.ts

@@ -0,0 +1,30 @@
+import { type FactoryProvider, type ModuleMetadata } from '@nestjs/common';
+
+import { type E2BDriverOptions } from './drivers/e2b.driver';
+import { type LocalDriverOptions } from './drivers/local.driver';
+
+export enum CodeInterpreterDriverType {
+  LOCAL = 'LOCAL',
+  E2B = 'E2B',
+}
+
+export type LocalDriverFactoryOptions = {
+  type: CodeInterpreterDriverType.LOCAL;
+  options: LocalDriverOptions;
+};
+
+export type E2BDriverFactoryOptions = {
+  type: CodeInterpreterDriverType.E2B;
+  options: E2BDriverOptions;
+};
+
+export type CodeInterpreterModuleOptions =
+  | LocalDriverFactoryOptions
+  | E2BDriverFactoryOptions;
+
+export type CodeInterpreterModuleAsyncOptions = {
+  useFactory: (
+    ...args: unknown[]
+  ) => CodeInterpreterModuleOptions | Promise<CodeInterpreterModuleOptions>;
+} & Pick<ModuleMetadata, 'imports'> &
+  Pick<FactoryProvider, 'inject'>;

packages/twenty-server/src/engine/core-modules/code-interpreter/code-interpreter.module.ts

@@ -0,0 +1,37 @@
+import { type DynamicModule, Global } from '@nestjs/common';
+
+import { CODE_INTERPRETER_DRIVER } from './code-interpreter.constants';
+import {
+  CodeInterpreterDriverType,
+  type CodeInterpreterModuleAsyncOptions,
+} from './code-interpreter.interface';
+import { CodeInterpreterService } from './code-interpreter.service';
+
+import { E2BDriver } from './drivers/e2b.driver';
+import { LocalDriver } from './drivers/local.driver';
+
+@Global()
+export class CodeInterpreterModule {
+  static forRootAsync(
+    options: CodeInterpreterModuleAsyncOptions,
+  ): DynamicModule {
+    const provider = {
+      provide: CODE_INTERPRETER_DRIVER,
+      useFactory: async (...args: unknown[]) => {
+        const config = await options.useFactory(...args);
+
+        return config.type === CodeInterpreterDriverType.LOCAL
+          ? new LocalDriver(config.options)
+          : new E2BDriver(config.options);
+      },
+      inject: options.inject ?? [],
+    };
+
+    return {
+      module: CodeInterpreterModule,
+      imports: options.imports ?? [],
+      providers: [CodeInterpreterService, provider],
+      exports: [CodeInterpreterService],
+    };
+  }
+}

packages/twenty-server/src/engine/core-modules/code-interpreter/code-interpreter.service.ts

@@ -0,0 +1,27 @@
+import { Inject, Injectable } from '@nestjs/common';
+
+import { CODE_INTERPRETER_DRIVER } from './code-interpreter.constants';
+
+import {
+  type CodeExecutionResult,
+  type CodeInterpreterDriver,
+  type ExecutionContext,
+  type InputFile,
+  type StreamCallbacks,
+} from './drivers/interfaces/code-interpreter-driver.interface';
+
+@Injectable()
+export class CodeInterpreterService implements CodeInterpreterDriver {
+  constructor(
+    @Inject(CODE_INTERPRETER_DRIVER) private driver: CodeInterpreterDriver,
+  ) {}
+
+  execute(
+    code: string,
+    files?: InputFile[],
+    context?: ExecutionContext,
+    callbacks?: StreamCallbacks,
+  ): Promise<CodeExecutionResult> {
+    return this.driver.execute(code, files, context, callbacks);
+  }
+}

packages/twenty-server/src/engine/core-modules/code-interpreter/drivers/e2b.driver.ts

@@ -0,0 +1,147 @@
+import { promises as fs } from 'fs';
+import { join } from 'path';
+
+import { Sandbox } from '@e2b/code-interpreter';
+
+import { DEFAULT_CODE_INTERPRETER_TIMEOUT_MS } from 'src/engine/core-modules/code-interpreter/code-interpreter.constants';
+import { getMimeType } from 'src/engine/core-modules/code-interpreter/utils/get-mime-type.util';
+
+import {
+  type CodeExecutionResult,
+  type CodeInterpreterDriver,
+  type ExecutionContext,
+  type InputFile,
+  type OutputFile,
+  type StreamCallbacks,
+} from './interfaces/code-interpreter-driver.interface';
+
+export type E2BDriverOptions = {
+  apiKey: string;
+  timeoutMs?: number;
+};
+
+const SANDBOX_SCRIPTS_PATH = join(__dirname, '..', 'sandbox-scripts');
+
+async function uploadDirectoryToSandbox(
+  sbx: Sandbox,
+  localPath: string,
+  remotePath: string,
+) {
+  const entries = await fs.readdir(localPath, { withFileTypes: true });
+
+  for (const entry of entries) {
+    const localEntryPath = join(localPath, entry.name);
+    const remoteEntryPath = `${remotePath}/${entry.name}`;
+
+    if (entry.isDirectory()) {
+      await uploadDirectoryToSandbox(sbx, localEntryPath, remoteEntryPath);
+    } else {
+      const content = await fs.readFile(localEntryPath);
+      const arrayBuffer = new Uint8Array(content).buffer;
+
+      await sbx.files.write(remoteEntryPath, arrayBuffer);
+    }
+  }
+}
+
+export class E2BDriver implements CodeInterpreterDriver {
+  constructor(private options: E2BDriverOptions) {}
+
+  async execute(
+    code: string,
+    files?: InputFile[],
+    context?: ExecutionContext,
+    callbacks?: StreamCallbacks,
+  ): Promise<CodeExecutionResult> {
+    const sbx = await Sandbox.create({
+      apiKey: this.options.apiKey,
+      timeoutMs: this.options.timeoutMs ?? DEFAULT_CODE_INTERPRETER_TIMEOUT_MS,
+    });
+
+    try {
+      // Upload pre-installed scripts to sandbox
+      try {
+        await uploadDirectoryToSandbox(
+          sbx,
+          SANDBOX_SCRIPTS_PATH,
+          '/home/user/scripts',
+        );
+      } catch {
+        // Scripts directory might not exist
+      }
+
+      for (const file of files ?? []) {
+        const arrayBuffer = new Uint8Array(file.content).buffer;
+
+        await sbx.files.write(`/home/user/${file.filename}`, arrayBuffer);
+      }
+
+      const envSetup = context?.env
+        ? `import os\n${Object.entries(context.env)
+            .map(([key, value]) => {
+              const escapedValue = value
+                .replace(/\\/g, '\\\\')
+                .replace(/'/g, "\\'")
+                .replace(/\n/g, '\\n')
+                .replace(/\r/g, '\\r');
+
+              return `os.environ['${key}'] = '${escapedValue}'`;
+            })
+            .join('\n')}\n\n`
+        : '';
+
+      const outputFiles: OutputFile[] = [];
+      let chartCounter = 0;
+
+      const execution = await sbx.runCode(envSetup + code, {
+        onStdout: (data) => callbacks?.onStdout?.(data.line),
+        onStderr: (data) => callbacks?.onStderr?.(data.line),
+        onResult: (result) => {
+          if (result.png) {
+            const outputFile: OutputFile = {
+              filename: `chart-${chartCounter++}.png`,
+              content: Buffer.from(result.png, 'base64'),
+              mimeType: 'image/png',
+            };
+
+            outputFiles.push(outputFile);
+            callbacks?.onResult?.(outputFile);
+          }
+        },
+      });
+
+      try {
+        const outputDir = await sbx.files.list('/home/user/output');
+
+        for (const file of outputDir) {
+          if (file.type === 'file') {
+            const content = await sbx.files.read(
+              `/home/user/output/${file.name}`,
+            );
+
+            const outputFile: OutputFile = {
+              filename: file.name,
+              content: Buffer.from(content),
+              mimeType: getMimeType(file.name),
+            };
+
+            outputFiles.push(outputFile);
+            callbacks?.onResult?.(outputFile);
+          }
+        }
+      } catch {
+        // Output directory doesn't exist - that's fine
+      }
+
+      return {
+        stdout: execution.logs.stdout.join('\n'),
+        stderr: execution.logs.stderr.join('\n'),
+        exitCode: execution.error ? 1 : 0,
+        files: outputFiles,
+        error: execution.error?.value,
+      };
+    } finally {
+      await sbx.kill();
+    }
+  }
+}

packages/twenty-server/src/engine/core-modules/code-interpreter/drivers/interfaces/code-interpreter-driver.interface.ts

@@ -0,0 +1,38 @@
+export type InputFile = {
+  filename: string;
+  content: Buffer;
+  mimeType: string;
+};
+
+export type OutputFile = {
+  filename: string;
+  content: Buffer;
+  mimeType: string;
+};
+
+export type CodeExecutionResult = {
+  stdout: string;
+  stderr: string;
+  exitCode: number;
+  files: OutputFile[];
+  error?: string;
+};
+
+export type ExecutionContext = {
+  env?: Record<string, string>;
+};
+
+export type StreamCallbacks = {
+  onStdout?: (line: string) => void;
+  onStderr?: (line: string) => void;
+  onResult?: (result: OutputFile) => void;
+};
+
+export interface CodeInterpreterDriver {
+  execute(
+    code: string,
+    files?: InputFile[],
+    context?: ExecutionContext,
+    callbacks?: StreamCallbacks,
+  ): Promise<CodeExecutionResult>;
+}

packages/twenty-server/src/engine/core-modules/code-interpreter/drivers/local.driver.ts

@@ -0,0 +1,212 @@
+import { spawn } from 'child_process';
+import { promises as fs } from 'fs';
+import { tmpdir } from 'os';
+import { basename, join } from 'path';
+
+import { DEFAULT_CODE_INTERPRETER_TIMEOUT_MS } from 'src/engine/core-modules/code-interpreter/code-interpreter.constants';
+import { getMimeType } from 'src/engine/core-modules/code-interpreter/utils/get-mime-type.util';
+
+import {
+  type CodeExecutionResult,
+  type CodeInterpreterDriver,
+  type ExecutionContext,
+  type InputFile,
+  type OutputFile,
+  type StreamCallbacks,
+} from './interfaces/code-interpreter-driver.interface';
+
+export type LocalDriverOptions = {
+  timeoutMs?: number;
+};
+
+const SANDBOX_SCRIPTS_PATH = join(__dirname, '..', 'sandbox-scripts');
+
+async function copyDirectoryRecursive(src: string, dest: string) {
+  await fs.mkdir(dest, { recursive: true });
+
+  const entries = await fs.readdir(src, { withFileTypes: true });
+
+  for (const entry of entries) {
+    const srcPath = join(src, entry.name);
+    const destPath = join(dest, entry.name);
+
+    if (entry.isDirectory()) {
+      await copyDirectoryRecursive(srcPath, destPath);
+    } else {
+      await fs.copyFile(srcPath, destPath);
+    }
+  }
+}
+
+// WARNING: This driver is UNSAFE and should only be used for development.
+// It executes arbitrary Python code on the server without any sandboxing.
+export class LocalDriver implements CodeInterpreterDriver {
+  constructor(private options: LocalDriverOptions = {}) {}
+
+  async execute(
+    code: string,
+    files?: InputFile[],
+    context?: ExecutionContext,
+    callbacks?: StreamCallbacks,
+  ): Promise<CodeExecutionResult> {
+    const workDir = await fs.mkdtemp(join(tmpdir(), 'code-interpreter-'));
+    const outputDir = join(workDir, 'output');
+    const scriptsDir = join(workDir, 'scripts');
+
+    await fs.mkdir(outputDir);
+
+    // Copy pre-installed scripts to sandbox
+    try {
+      await copyDirectoryRecursive(SANDBOX_SCRIPTS_PATH, scriptsDir);
+    } catch {
+      // Scripts directory might not exist in dev environment
+    }
+
+    try {
+      for (const file of files ?? []) {
+        const safeFilename = basename(file.filename);
+
+        await fs.writeFile(join(workDir, safeFilename), file.content);
+      }
+
+      // Rewrite E2B-style paths to local paths for compatibility
+      const rewrittenCode = code
+        .replace(/\/home\/user\/scripts\//g, `${scriptsDir}/`)
+        .replace(/\/home\/user\/scripts/g, scriptsDir)
+        .replace(/\/home\/user\/output\//g, `${outputDir}/`)
+        .replace(/\/home\/user\/output/g, outputDir)
+        .replace(/\/home\/user\//g, `${workDir}/`)
+        .replace(/\/home\/user/g, workDir);
+
+      const scriptPath = join(workDir, 'script.py');
+
+      await fs.writeFile(scriptPath, rewrittenCode);
+
+      const timeoutMs =
+        this.options.timeoutMs ?? DEFAULT_CODE_INTERPRETER_TIMEOUT_MS;
+
+      const { stdout, stderr, exitCode, error } = await this.runPythonScript(
+        scriptPath,
+        workDir,
+        outputDir,
+        context?.env,
+        timeoutMs,
+        callbacks,
+      );
+
+      const outputFiles: OutputFile[] = [];
+
+      try {
+        const outputEntries = await fs.readdir(outputDir, {
+          withFileTypes: true,
+        });
+
+        for (const entry of outputEntries) {
+          if (entry.isFile()) {
+            const content = await fs.readFile(join(outputDir, entry.name));
+            const outputFile: OutputFile = {
+              filename: entry.name,
+              content,
+              mimeType: getMimeType(entry.name),
+            };
+
+            outputFiles.push(outputFile);
+            callbacks?.onResult?.(outputFile);
+          }
+        }
+      } catch {
+        // Output directory might be empty or not exist
+      }
+
+      return {
+        stdout,
+        stderr,
+        exitCode,
+        files: outputFiles,
+        error,
+      };
+    } finally {
+      await fs.rm(workDir, { recursive: true, force: true });
+    }
+  }
+
+  private runPythonScript(
+    scriptPath: string,
+    workDir: string,
+    outputDir: string,
+    env?: Record<string, string>,
+    timeoutMs?: number,
+    callbacks?: StreamCallbacks,
+  ): Promise<{
+    stdout: string;
+    stderr: string;
+    exitCode: number;
+    error?: string;
+  }> {
+    return new Promise((resolve) => {
+      const child = spawn('python3', [scriptPath], {
+        cwd: workDir,
+        env: {
+          ...process.env,
+          OUTPUT_DIR: outputDir,
+          ...env,
+        },
+      });
+
+      let stdout = '';
+      let stderr = '';
+      let killed = false;
+
+      const timeout = setTimeout(() => {
+        killed = true;
+        child.kill('SIGKILL');
+      }, timeoutMs ?? DEFAULT_CODE_INTERPRETER_TIMEOUT_MS);
+
+      child.stdout.on('data', (data: Buffer) => {
+        const text = data.toString();
+
+        stdout += text;
+        const lines = text.split('\n');
+
+        for (const line of lines) {
+          if (line) {
+            callbacks?.onStdout?.(line);
+          }
+        }
+      });
+
+      child.stderr.on('data', (data: Buffer) => {
+        const text = data.toString();
+
+        stderr += text;
+        const lines = text.split('\n');
+
+        for (const line of lines) {
+          if (line) {
+            callbacks?.onStderr?.(line);
+          }
+        }
+      });
+
+      child.on('close', (code) => {
+        clearTimeout(timeout);
+        resolve({
+          stdout,
+          stderr,
+          exitCode: code ?? 0,
+          error: killed ? 'Process timed out' : undefined,
+        });
+      });
+
+      child.on('error', (err) => {
+        clearTimeout(timeout);
+        resolve({
+          stdout,
+          stderr,
+          exitCode: 1,
+          error: err.message,
+        });
+      });
+    });
+  }
+}

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/pack.py

@@ -0,0 +1,159 @@
+#!/usr/bin/env python3
+"""
+Tool to pack a directory into a .docx, .pptx, or .xlsx file with XML formatting undone.
+
+Example usage:
+    python pack.py <input_directory> <office_file> [--force]
+"""
+
+import argparse
+import shutil
+import subprocess
+import sys
+import tempfile
+import defusedxml.minidom
+import zipfile
+from pathlib import Path
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Pack a directory into an Office file")
+    parser.add_argument("input_directory", help="Unpacked Office document directory")
+    parser.add_argument("output_file", help="Output Office file (.docx/.pptx/.xlsx)")
+    parser.add_argument("--force", action="store_true", help="Skip validation")
+    args = parser.parse_args()
+
+    try:
+        success = pack_document(
+            args.input_directory, args.output_file, validate=not args.force
+        )
+
+        # Show warning if validation was skipped
+        if args.force:
+            print("Warning: Skipped validation, file may be corrupt", file=sys.stderr)
+        # Exit with error if validation failed
+        elif not success:
+            print("Contents would produce a corrupt file.", file=sys.stderr)
+            print("Please validate XML before repacking.", file=sys.stderr)
+            print("Use --force to skip validation and pack anyway.", file=sys.stderr)
+            sys.exit(1)
+
+    except ValueError as e:
+        sys.exit(f"Error: {e}")
+
+
+def pack_document(input_dir, output_file, validate=False):
+    """Pack a directory into an Office file (.docx/.pptx/.xlsx).
+
+    Args:
+        input_dir: Path to unpacked Office document directory
+        output_file: Path to output Office file
+        validate: If True, validates with soffice (default: False)
+
+    Returns:
+        bool: True if successful, False if validation failed
+    """
+    input_dir = Path(input_dir)
+    output_file = Path(output_file)
+
+    if not input_dir.is_dir():
+        raise ValueError(f"{input_dir} is not a directory")
+    if output_file.suffix.lower() not in {".docx", ".pptx", ".xlsx"}:
+        raise ValueError(f"{output_file} must be a .docx, .pptx, or .xlsx file")
+
+    # Work in temporary directory to avoid modifying original
+    with tempfile.TemporaryDirectory() as temp_dir:
+        temp_content_dir = Path(temp_dir) / "content"
+        shutil.copytree(input_dir, temp_content_dir)
+
+        # Process XML files to remove pretty-printing whitespace
+        for pattern in ["*.xml", "*.rels"]:
+            for xml_file in temp_content_dir.rglob(pattern):
+                condense_xml(xml_file)
+
+        # Create final Office file as zip archive
+        output_file.parent.mkdir(parents=True, exist_ok=True)
+        with zipfile.ZipFile(output_file, "w", zipfile.ZIP_DEFLATED) as zf:
+            for f in temp_content_dir.rglob("*"):
+                if f.is_file():
+                    zf.write(f, f.relative_to(temp_content_dir))
+
+        # Validate if requested
+        if validate:
+            if not validate_document(output_file):
+                output_file.unlink()  # Delete the corrupt file
+                return False
+
+    return True
+
+
+def validate_document(doc_path):
+    """Validate document by converting to HTML with soffice."""
+    # Determine the correct filter based on file extension
+    match doc_path.suffix.lower():
+        case ".docx":
+            filter_name = "html:HTML"
+        case ".pptx":
+            filter_name = "html:impress_html_Export"
+        case ".xlsx":
+            filter_name = "html:HTML (StarCalc)"
+
+    with tempfile.TemporaryDirectory() as temp_dir:
+        try:
+            result = subprocess.run(
+                [
+                    "soffice",
+                    "--headless",
+                    "--convert-to",
+                    filter_name,
+                    "--outdir",
+                    temp_dir,
+                    str(doc_path),
+                ],
+                capture_output=True,
+                timeout=10,
+                text=True,
+            )
+            if not (Path(temp_dir) / f"{doc_path.stem}.html").exists():
+                error_msg = result.stderr.strip() or "Document validation failed"
+                print(f"Validation error: {error_msg}", file=sys.stderr)
+                return False
+            return True
+        except FileNotFoundError:
+            print("Warning: soffice not found. Skipping validation.", file=sys.stderr)
+            return True
+        except subprocess.TimeoutExpired:
+            print("Validation error: Timeout during conversion", file=sys.stderr)
+            return False
+        except Exception as e:
+            print(f"Validation error: {e}", file=sys.stderr)
+            return False
+
+
+def condense_xml(xml_file):
+    """Strip unnecessary whitespace and remove comments."""
+    with open(xml_file, "r", encoding="utf-8") as f:
+        dom = defusedxml.minidom.parse(f)
+
+    # Process each element to remove whitespace and comments
+    for element in dom.getElementsByTagName("*"):
+        # Skip w:t elements and their processing
+        if element.tagName.endswith(":t"):
+            continue
+
+        # Remove whitespace-only text nodes and comment nodes
+        for child in list(element.childNodes):
+            if (
+                child.nodeType == child.TEXT_NODE
+                and child.nodeValue
+                and child.nodeValue.strip() == ""
+            ) or child.nodeType == child.COMMENT_NODE:
+                element.removeChild(child)
+
+    # Write back the condensed XML
+    with open(xml_file, "wb") as f:
+        f.write(dom.toxml(encoding="UTF-8"))
+
+
+if __name__ == "__main__":
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/unpack.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python3
+"""Unpack and format XML contents of Office files (.docx, .pptx, .xlsx)"""
+
+import random
+import sys
+import defusedxml.minidom
+import zipfile
+from pathlib import Path
+
+# Get command line arguments
+assert len(sys.argv) == 3, "Usage: python unpack.py <office_file> <output_dir>"
+input_file, output_dir = sys.argv[1], sys.argv[2]
+
+# Extract and format
+output_path = Path(output_dir)
+output_path.mkdir(parents=True, exist_ok=True)
+zipfile.ZipFile(input_file).extractall(output_path)
+
+# Pretty print all XML files
+xml_files = list(output_path.rglob("*.xml")) + list(output_path.rglob("*.rels"))
+for xml_file in xml_files:
+    content = xml_file.read_text(encoding="utf-8")
+    dom = defusedxml.minidom.parseString(content)
+    xml_file.write_bytes(dom.toprettyxml(indent="  ", encoding="ascii"))
+
+# For .docx files, suggest an RSID for tracked changes
+if input_file.endswith(".docx"):
+    suggested_rsid = "".join(random.choices("0123456789ABCDEF", k=8))
+    print(f"Suggested RSID for edit session: {suggested_rsid}")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validate.py

@@ -0,0 +1,69 @@
+#!/usr/bin/env python3
+"""
+Command line tool to validate Office document XML files against XSD schemas and tracked changes.
+
+Usage:
+    python validate.py <dir> --original <original_file>
+"""
+
+import argparse
+import sys
+from pathlib import Path
+
+from validation import DOCXSchemaValidator, PPTXSchemaValidator, RedliningValidator
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Validate Office document XML files")
+    parser.add_argument(
+        "unpacked_dir",
+        help="Path to unpacked Office document directory",
+    )
+    parser.add_argument(
+        "--original",
+        required=True,
+        help="Path to original file (.docx/.pptx/.xlsx)",
+    )
+    parser.add_argument(
+        "-v",
+        "--verbose",
+        action="store_true",
+        help="Enable verbose output",
+    )
+    args = parser.parse_args()
+
+    # Validate paths
+    unpacked_dir = Path(args.unpacked_dir)
+    original_file = Path(args.original)
+    file_extension = original_file.suffix.lower()
+    assert unpacked_dir.is_dir(), f"Error: {unpacked_dir} is not a directory"
+    assert original_file.is_file(), f"Error: {original_file} is not a file"
+    assert file_extension in [".docx", ".pptx", ".xlsx"], (
+        f"Error: {original_file} must be a .docx, .pptx, or .xlsx file"
+    )
+
+    # Run validations
+    match file_extension:
+        case ".docx":
+            validators = [DOCXSchemaValidator, RedliningValidator]
+        case ".pptx":
+            validators = [PPTXSchemaValidator]
+        case _:
+            print(f"Error: Validation not supported for file type {file_extension}")
+            sys.exit(1)
+
+    # Run validators
+    success = True
+    for V in validators:
+        validator = V(unpacked_dir, original_file, verbose=args.verbose)
+        if not validator.validate():
+            success = False
+
+    if success:
+        print("All validations PASSED!")
+
+    sys.exit(0 if success else 1)
+
+
+if __name__ == "__main__":
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validation/__init__.py

@@ -0,0 +1,15 @@
+"""
+Validation modules for Word document processing.
+"""
+
+from .base import BaseSchemaValidator
+from .docx import DOCXSchemaValidator
+from .pptx import PPTXSchemaValidator
+from .redlining import RedliningValidator
+
+__all__ = [
+    "BaseSchemaValidator",
+    "DOCXSchemaValidator",
+    "PPTXSchemaValidator",
+    "RedliningValidator",
+]

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validation/base.py

@@ -0,0 +1,951 @@
+"""
+Base validator with common validation logic for document files.
+"""
+
+import re
+from pathlib import Path
+
+import lxml.etree
+
+
+class BaseSchemaValidator:
+    """Base validator with common validation logic for document files."""
+
+    # Elements whose 'id' attributes must be unique within their file
+    # Format: element_name -> (attribute_name, scope)
+    # scope can be 'file' (unique within file) or 'global' (unique across all files)
+    UNIQUE_ID_REQUIREMENTS = {
+        # Word elements
+        "comment": ("id", "file"),  # Comment IDs in comments.xml
+        "commentrangestart": ("id", "file"),  # Must match comment IDs
+        "commentrangeend": ("id", "file"),  # Must match comment IDs
+        "bookmarkstart": ("id", "file"),  # Bookmark start IDs
+        "bookmarkend": ("id", "file"),  # Bookmark end IDs
+        # Note: ins and del (track changes) can share IDs when part of same revision
+        # PowerPoint elements
+        "sldid": ("id", "file"),  # Slide IDs in presentation.xml
+        "sldmasterid": ("id", "global"),  # Slide master IDs must be globally unique
+        "sldlayoutid": ("id", "global"),  # Slide layout IDs must be globally unique
+        "cm": ("authorid", "file"),  # Comment author IDs
+        # Excel elements
+        "sheet": ("sheetid", "file"),  # Sheet IDs in workbook.xml
+        "definedname": ("id", "file"),  # Named range IDs
+        # Drawing/Shape elements (all formats)
+        "cxnsp": ("id", "file"),  # Connection shape IDs
+        "sp": ("id", "file"),  # Shape IDs
+        "pic": ("id", "file"),  # Picture IDs
+        "grpsp": ("id", "file"),  # Group shape IDs
+    }
+
+    # Mapping of element names to expected relationship types
+    # Subclasses should override this with format-specific mappings
+    ELEMENT_RELATIONSHIP_TYPES = {}
+
+    # Unified schema mappings for all Office document types
+    SCHEMA_MAPPINGS = {
+        # Document type specific schemas
+        "word": "ISO-IEC29500-4_2016/wml.xsd",  # Word documents
+        "ppt": "ISO-IEC29500-4_2016/pml.xsd",  # PowerPoint presentations
+        "xl": "ISO-IEC29500-4_2016/sml.xsd",  # Excel spreadsheets
+        # Common file types
+        "[Content_Types].xml": "ecma/fouth-edition/opc-contentTypes.xsd",
+        "app.xml": "ISO-IEC29500-4_2016/shared-documentPropertiesExtended.xsd",
+        "core.xml": "ecma/fouth-edition/opc-coreProperties.xsd",
+        "custom.xml": "ISO-IEC29500-4_2016/shared-documentPropertiesCustom.xsd",
+        ".rels": "ecma/fouth-edition/opc-relationships.xsd",
+        # Word-specific files
+        "people.xml": "microsoft/wml-2012.xsd",
+        "commentsIds.xml": "microsoft/wml-cid-2016.xsd",
+        "commentsExtensible.xml": "microsoft/wml-cex-2018.xsd",
+        "commentsExtended.xml": "microsoft/wml-2012.xsd",
+        # Chart files (common across document types)
+        "chart": "ISO-IEC29500-4_2016/dml-chart.xsd",
+        # Theme files (common across document types)
+        "theme": "ISO-IEC29500-4_2016/dml-main.xsd",
+        # Drawing and media files
+        "drawing": "ISO-IEC29500-4_2016/dml-main.xsd",
+    }
+
+    # Unified namespace constants
+    MC_NAMESPACE = "http://schemas.openxmlformats.org/markup-compatibility/2006"
+    XML_NAMESPACE = "http://www.w3.org/XML/1998/namespace"
+
+    # Common OOXML namespaces used across validators
+    PACKAGE_RELATIONSHIPS_NAMESPACE = (
+        "http://schemas.openxmlformats.org/package/2006/relationships"
+    )
+    OFFICE_RELATIONSHIPS_NAMESPACE = (
+        "http://schemas.openxmlformats.org/officeDocument/2006/relationships"
+    )
+    CONTENT_TYPES_NAMESPACE = (
+        "http://schemas.openxmlformats.org/package/2006/content-types"
+    )
+
+    # Folders where we should clean ignorable namespaces
+    MAIN_CONTENT_FOLDERS = {"word", "ppt", "xl"}
+
+    # All allowed OOXML namespaces (superset of all document types)
+    OOXML_NAMESPACES = {
+        "http://schemas.openxmlformats.org/officeDocument/2006/math",
+        "http://schemas.openxmlformats.org/officeDocument/2006/relationships",
+        "http://schemas.openxmlformats.org/schemaLibrary/2006/main",
+        "http://schemas.openxmlformats.org/drawingml/2006/main",
+        "http://schemas.openxmlformats.org/drawingml/2006/chart",
+        "http://schemas.openxmlformats.org/drawingml/2006/chartDrawing",
+        "http://schemas.openxmlformats.org/drawingml/2006/diagram",
+        "http://schemas.openxmlformats.org/drawingml/2006/picture",
+        "http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing",
+        "http://schemas.openxmlformats.org/drawingml/2006/wordprocessingDrawing",
+        "http://schemas.openxmlformats.org/wordprocessingml/2006/main",
+        "http://schemas.openxmlformats.org/presentationml/2006/main",
+        "http://schemas.openxmlformats.org/spreadsheetml/2006/main",
+        "http://schemas.openxmlformats.org/officeDocument/2006/sharedTypes",
+        "http://www.w3.org/XML/1998/namespace",
+    }
+
+    def __init__(self, unpacked_dir, original_file, verbose=False):
+        self.unpacked_dir = Path(unpacked_dir).resolve()
+        self.original_file = Path(original_file)
+        self.verbose = verbose
+
+        # Set schemas directory
+        self.schemas_dir = Path(__file__).parent.parent.parent / "schemas"
+
+        # Get all XML and .rels files
+        patterns = ["*.xml", "*.rels"]
+        self.xml_files = [
+            f for pattern in patterns for f in self.unpacked_dir.rglob(pattern)
+        ]
+
+        if not self.xml_files:
+            print(f"Warning: No XML files found in {self.unpacked_dir}")
+
+    def validate(self):
+        """Run all validation checks and return True if all pass."""
+        raise NotImplementedError("Subclasses must implement the validate method")
+
+    def validate_xml(self):
+        """Validate that all XML files are well-formed."""
+        errors = []
+
+        for xml_file in self.xml_files:
+            try:
+                # Try to parse the XML file
+                lxml.etree.parse(str(xml_file))
+            except lxml.etree.XMLSyntaxError as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                    f"Line {e.lineno}: {e.msg}"
+                )
+            except Exception as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                    f"Unexpected error: {str(e)}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} XML violations:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All XML files are well-formed")
+            return True
+
+    def validate_namespaces(self):
+        """Validate that namespace prefixes in Ignorable attributes are declared."""
+        errors = []
+
+        for xml_file in self.xml_files:
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+                declared = set(root.nsmap.keys()) - {None}  # Exclude default namespace
+
+                for attr_val in [
+                    v for k, v in root.attrib.items() if k.endswith("Ignorable")
+                ]:
+                    undeclared = set(attr_val.split()) - declared
+                    errors.extend(
+                        f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                        f"Namespace '{ns}' in Ignorable but not declared"
+                        for ns in undeclared
+                    )
+            except lxml.etree.XMLSyntaxError:
+                continue
+
+        if errors:
+            print(f"FAILED - {len(errors)} namespace issues:")
+            for error in errors:
+                print(error)
+            return False
+        if self.verbose:
+            print("PASSED - All namespace prefixes properly declared")
+        return True
+
+    def validate_unique_ids(self):
+        """Validate that specific IDs are unique according to OOXML requirements."""
+        errors = []
+        global_ids = {}  # Track globally unique IDs across all files
+
+        for xml_file in self.xml_files:
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+                file_ids = {}  # Track IDs that must be unique within this file
+
+                # Remove all mc:AlternateContent elements from the tree
+                mc_elements = root.xpath(
+                    ".//mc:AlternateContent", namespaces={"mc": self.MC_NAMESPACE}
+                )
+                for elem in mc_elements:
+                    elem.getparent().remove(elem)
+
+                # Now check IDs in the cleaned tree
+                for elem in root.iter():
+                    # Get the element name without namespace
+                    tag = (
+                        elem.tag.split("}")[-1].lower()
+                        if "}" in elem.tag
+                        else elem.tag.lower()
+                    )
+
+                    # Check if this element type has ID uniqueness requirements
+                    if tag in self.UNIQUE_ID_REQUIREMENTS:
+                        attr_name, scope = self.UNIQUE_ID_REQUIREMENTS[tag]
+
+                        # Look for the specified attribute
+                        id_value = None
+                        for attr, value in elem.attrib.items():
+                            attr_local = (
+                                attr.split("}")[-1].lower()
+                                if "}" in attr
+                                else attr.lower()
+                            )
+                            if attr_local == attr_name:
+                                id_value = value
+                                break
+
+                        if id_value is not None:
+                            if scope == "global":
+                                # Check global uniqueness
+                                if id_value in global_ids:
+                                    prev_file, prev_line, prev_tag = global_ids[
+                                        id_value
+                                    ]
+                                    errors.append(
+                                        f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                                        f"Line {elem.sourceline}: Global ID '{id_value}' in <{tag}> "
+                                        f"already used in {prev_file} at line {prev_line} in <{prev_tag}>"
+                                    )
+                                else:
+                                    global_ids[id_value] = (
+                                        xml_file.relative_to(self.unpacked_dir),
+                                        elem.sourceline,
+                                        tag,
+                                    )
+                            elif scope == "file":
+                                # Check file-level uniqueness
+                                key = (tag, attr_name)
+                                if key not in file_ids:
+                                    file_ids[key] = {}
+
+                                if id_value in file_ids[key]:
+                                    prev_line = file_ids[key][id_value]
+                                    errors.append(
+                                        f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                                        f"Line {elem.sourceline}: Duplicate {attr_name}='{id_value}' in <{tag}> "
+                                        f"(first occurrence at line {prev_line})"
+                                    )
+                                else:
+                                    file_ids[key][id_value] = elem.sourceline
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} ID uniqueness violations:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All required IDs are unique")
+            return True
+
+    def validate_file_references(self):
+        """
+        Validate that all .rels files properly reference files and that all files are referenced.
+        """
+        errors = []
+
+        # Find all .rels files
+        rels_files = list(self.unpacked_dir.rglob("*.rels"))
+
+        if not rels_files:
+            if self.verbose:
+                print("PASSED - No .rels files found")
+            return True
+
+        # Get all files in the unpacked directory (excluding reference files)
+        all_files = []
+        for file_path in self.unpacked_dir.rglob("*"):
+            if (
+                file_path.is_file()
+                and file_path.name != "[Content_Types].xml"
+                and not file_path.name.endswith(".rels")
+            ):  # This file is not referenced by .rels
+                all_files.append(file_path.resolve())
+
+        # Track all files that are referenced by any .rels file
+        all_referenced_files = set()
+
+        if self.verbose:
+            print(
+                f"Found {len(rels_files)} .rels files and {len(all_files)} target files"
+            )
+
+        # Check each .rels file
+        for rels_file in rels_files:
+            try:
+                # Parse relationships file
+                rels_root = lxml.etree.parse(str(rels_file)).getroot()
+
+                # Get the directory where this .rels file is located
+                rels_dir = rels_file.parent
+
+                # Find all relationships and their targets
+                referenced_files = set()
+                broken_refs = []
+
+                for rel in rels_root.findall(
+                    ".//ns:Relationship",
+                    namespaces={"ns": self.PACKAGE_RELATIONSHIPS_NAMESPACE},
+                ):
+                    target = rel.get("Target")
+                    if target and not target.startswith(
+                        ("http", "mailto:")
+                    ):  # Skip external URLs
+                        # Resolve the target path relative to the .rels file location
+                        if rels_file.name == ".rels":
+                            # Root .rels file - targets are relative to unpacked_dir
+                            target_path = self.unpacked_dir / target
+                        else:
+                            # Other .rels files - targets are relative to their parent's parent
+                            # e.g., word/_rels/document.xml.rels -> targets relative to word/
+                            base_dir = rels_dir.parent
+                            target_path = base_dir / target
+
+                        # Normalize the path and check if it exists
+                        try:
+                            target_path = target_path.resolve()
+                            if target_path.exists() and target_path.is_file():
+                                referenced_files.add(target_path)
+                                all_referenced_files.add(target_path)
+                            else:
+                                broken_refs.append((target, rel.sourceline))
+                        except (OSError, ValueError):
+                            broken_refs.append((target, rel.sourceline))
+
+                # Report broken references
+                if broken_refs:
+                    rel_path = rels_file.relative_to(self.unpacked_dir)
+                    for broken_ref, line_num in broken_refs:
+                        errors.append(
+                            f"  {rel_path}: Line {line_num}: Broken reference to {broken_ref}"
+                        )
+
+            except Exception as e:
+                rel_path = rels_file.relative_to(self.unpacked_dir)
+                errors.append(f"  Error parsing {rel_path}: {e}")
+
+        # Check for unreferenced files (files that exist but are not referenced anywhere)
+        unreferenced_files = set(all_files) - all_referenced_files
+
+        if unreferenced_files:
+            for unref_file in sorted(unreferenced_files):
+                unref_rel_path = unref_file.relative_to(self.unpacked_dir)
+                errors.append(f"  Unreferenced file: {unref_rel_path}")
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} relationship validation errors:")
+            for error in errors:
+                print(error)
+            print(
+                "CRITICAL: These errors will cause the document to appear corrupt. "
+                + "Broken references MUST be fixed, "
+                + "and unreferenced files MUST be referenced or removed."
+            )
+            return False
+        else:
+            if self.verbose:
+                print(
+                    "PASSED - All references are valid and all files are properly referenced"
+                )
+            return True
+
+    def validate_all_relationship_ids(self):
+        """
+        Validate that all r:id attributes in XML files reference existing IDs
+        in their corresponding .rels files, and optionally validate relationship types.
+        """
+        import lxml.etree
+
+        errors = []
+
+        # Process each XML file that might contain r:id references
+        for xml_file in self.xml_files:
+            # Skip .rels files themselves
+            if xml_file.suffix == ".rels":
+                continue
+
+            # Determine the corresponding .rels file
+            # For dir/file.xml, it's dir/_rels/file.xml.rels
+            rels_dir = xml_file.parent / "_rels"
+            rels_file = rels_dir / f"{xml_file.name}.rels"
+
+            # Skip if there's no corresponding .rels file (that's okay)
+            if not rels_file.exists():
+                continue
+
+            try:
+                # Parse the .rels file to get valid relationship IDs and their types
+                rels_root = lxml.etree.parse(str(rels_file)).getroot()
+                rid_to_type = {}
+
+                for rel in rels_root.findall(
+                    f".//{{{self.PACKAGE_RELATIONSHIPS_NAMESPACE}}}Relationship"
+                ):
+                    rid = rel.get("Id")
+                    rel_type = rel.get("Type", "")
+                    if rid:
+                        # Check for duplicate rIds
+                        if rid in rid_to_type:
+                            rels_rel_path = rels_file.relative_to(self.unpacked_dir)
+                            errors.append(
+                                f"  {rels_rel_path}: Line {rel.sourceline}: "
+                                f"Duplicate relationship ID '{rid}' (IDs must be unique)"
+                            )
+                        # Extract just the type name from the full URL
+                        type_name = (
+                            rel_type.split("/")[-1] if "/" in rel_type else rel_type
+                        )
+                        rid_to_type[rid] = type_name
+
+                # Parse the XML file to find all r:id references
+                xml_root = lxml.etree.parse(str(xml_file)).getroot()
+
+                # Find all elements with r:id attributes
+                for elem in xml_root.iter():
+                    # Check for r:id attribute (relationship ID)
+                    rid_attr = elem.get(f"{{{self.OFFICE_RELATIONSHIPS_NAMESPACE}}}id")
+                    if rid_attr:
+                        xml_rel_path = xml_file.relative_to(self.unpacked_dir)
+                        elem_name = (
+                            elem.tag.split("}")[-1] if "}" in elem.tag else elem.tag
+                        )
+
+                        # Check if the ID exists
+                        if rid_attr not in rid_to_type:
+                            errors.append(
+                                f"  {xml_rel_path}: Line {elem.sourceline}: "
+                                f"<{elem_name}> references non-existent relationship '{rid_attr}' "
+                                f"(valid IDs: {', '.join(sorted(rid_to_type.keys())[:5])}{'...' if len(rid_to_type) > 5 else ''})"
+                            )
+                        # Check if we have type expectations for this element
+                        elif self.ELEMENT_RELATIONSHIP_TYPES:
+                            expected_type = self._get_expected_relationship_type(
+                                elem_name
+                            )
+                            if expected_type:
+                                actual_type = rid_to_type[rid_attr]
+                                # Check if the actual type matches or contains the expected type
+                                if expected_type not in actual_type.lower():
+                                    errors.append(
+                                        f"  {xml_rel_path}: Line {elem.sourceline}: "
+                                        f"<{elem_name}> references '{rid_attr}' which points to '{actual_type}' "
+                                        f"but should point to a '{expected_type}' relationship"
+                                    )
+
+            except Exception as e:
+                xml_rel_path = xml_file.relative_to(self.unpacked_dir)
+                errors.append(f"  Error processing {xml_rel_path}: {e}")
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} relationship ID reference errors:")
+            for error in errors:
+                print(error)
+            print("\nThese ID mismatches will cause the document to appear corrupt!")
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All relationship ID references are valid")
+            return True
+
+    def _get_expected_relationship_type(self, element_name):
+        """
+        Get the expected relationship type for an element.
+        First checks the explicit mapping, then tries pattern detection.
+        """
+        # Normalize element name to lowercase
+        elem_lower = element_name.lower()
+
+        # Check explicit mapping first
+        if elem_lower in self.ELEMENT_RELATIONSHIP_TYPES:
+            return self.ELEMENT_RELATIONSHIP_TYPES[elem_lower]
+
+        # Try pattern detection for common patterns
+        # Pattern 1: Elements ending in "Id" often expect a relationship of the prefix type
+        if elem_lower.endswith("id") and len(elem_lower) > 2:
+            # e.g., "sldId" -> "sld", "sldMasterId" -> "sldMaster"
+            prefix = elem_lower[:-2]  # Remove "id"
+            # Check if this might be a compound like "sldMasterId"
+            if prefix.endswith("master"):
+                return prefix.lower()
+            elif prefix.endswith("layout"):
+                return prefix.lower()
+            else:
+                # Simple case like "sldId" -> "slide"
+                # Common transformations
+                if prefix == "sld":
+                    return "slide"
+                return prefix.lower()
+
+        # Pattern 2: Elements ending in "Reference" expect a relationship of the prefix type
+        if elem_lower.endswith("reference") and len(elem_lower) > 9:
+            prefix = elem_lower[:-9]  # Remove "reference"
+            return prefix.lower()
+
+        return None
+
+    def validate_content_types(self):
+        """Validate that all content files are properly declared in [Content_Types].xml."""
+        errors = []
+
+        # Find [Content_Types].xml file
+        content_types_file = self.unpacked_dir / "[Content_Types].xml"
+        if not content_types_file.exists():
+            print("FAILED - [Content_Types].xml file not found")
+            return False
+
+        try:
+            # Parse and get all declared parts and extensions
+            root = lxml.etree.parse(str(content_types_file)).getroot()
+            declared_parts = set()
+            declared_extensions = set()
+
+            # Get Override declarations (specific files)
+            for override in root.findall(
+                f".//{{{self.CONTENT_TYPES_NAMESPACE}}}Override"
+            ):
+                part_name = override.get("PartName")
+                if part_name is not None:
+                    declared_parts.add(part_name.lstrip("/"))
+
+            # Get Default declarations (by extension)
+            for default in root.findall(
+                f".//{{{self.CONTENT_TYPES_NAMESPACE}}}Default"
+            ):
+                extension = default.get("Extension")
+                if extension is not None:
+                    declared_extensions.add(extension.lower())
+
+            # Root elements that require content type declaration
+            declarable_roots = {
+                "sld",
+                "sldLayout",
+                "sldMaster",
+                "presentation",  # PowerPoint
+                "document",  # Word
+                "workbook",
+                "worksheet",  # Excel
+                "theme",  # Common
+            }
+
+            # Common media file extensions that should be declared
+            media_extensions = {
+                "png": "image/png",
+                "jpg": "image/jpeg",
+                "jpeg": "image/jpeg",
+                "gif": "image/gif",
+                "bmp": "image/bmp",
+                "tiff": "image/tiff",
+                "wmf": "image/x-wmf",
+                "emf": "image/x-emf",
+            }
+
+            # Get all files in the unpacked directory
+            all_files = list(self.unpacked_dir.rglob("*"))
+            all_files = [f for f in all_files if f.is_file()]
+
+            # Check all XML files for Override declarations
+            for xml_file in self.xml_files:
+                path_str = str(xml_file.relative_to(self.unpacked_dir)).replace(
+                    "\\", "/"
+                )
+
+                # Skip non-content files
+                if any(
+                    skip in path_str
+                    for skip in [".rels", "[Content_Types]", "docProps/", "_rels/"]
+                ):
+                    continue
+
+                try:
+                    root_tag = lxml.etree.parse(str(xml_file)).getroot().tag
+                    root_name = root_tag.split("}")[-1] if "}" in root_tag else root_tag
+
+                    if root_name in declarable_roots and path_str not in declared_parts:
+                        errors.append(
+                            f"  {path_str}: File with <{root_name}> root not declared in [Content_Types].xml"
+                        )
+
+                except Exception:
+                    continue  # Skip unparseable files
+
+            # Check all non-XML files for Default extension declarations
+            for file_path in all_files:
+                # Skip XML files and metadata files (already checked above)
+                if file_path.suffix.lower() in {".xml", ".rels"}:
+                    continue
+                if file_path.name == "[Content_Types].xml":
+                    continue
+                if "_rels" in file_path.parts or "docProps" in file_path.parts:
+                    continue
+
+                extension = file_path.suffix.lstrip(".").lower()
+                if extension and extension not in declared_extensions:
+                    # Check if it's a known media extension that should be declared
+                    if extension in media_extensions:
+                        relative_path = file_path.relative_to(self.unpacked_dir)
+                        errors.append(
+                            f'  {relative_path}: File with extension \'{extension}\' not declared in [Content_Types].xml - should add: <Default Extension="{extension}" ContentType="{media_extensions[extension]}"/>'
+                        )
+
+        except Exception as e:
+            errors.append(f"  Error parsing [Content_Types].xml: {e}")
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} content type declaration errors:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print(
+                    "PASSED - All content files are properly declared in [Content_Types].xml"
+                )
+            return True
+
+    def validate_file_against_xsd(self, xml_file, verbose=False):
+        """Validate a single XML file against XSD schema, comparing with original.
+
+        Args:
+            xml_file: Path to XML file to validate
+            verbose: Enable verbose output
+
+        Returns:
+            tuple: (is_valid, new_errors_set) where is_valid is True/False/None (skipped)
+        """
+        # Resolve both paths to handle symlinks
+        xml_file = Path(xml_file).resolve()
+        unpacked_dir = self.unpacked_dir.resolve()
+
+        # Validate current file
+        is_valid, current_errors = self._validate_single_file_xsd(
+            xml_file, unpacked_dir
+        )
+
+        if is_valid is None:
+            return None, set()  # Skipped
+        elif is_valid:
+            return True, set()  # Valid, no errors
+
+        # Get errors from original file for this specific file
+        original_errors = self._get_original_file_errors(xml_file)
+
+        # Compare with original (both are guaranteed to be sets here)
+        assert current_errors is not None
+        new_errors = current_errors - original_errors
+
+        if new_errors:
+            if verbose:
+                relative_path = xml_file.relative_to(unpacked_dir)
+                print(f"FAILED - {relative_path}: {len(new_errors)} new error(s)")
+                for error in list(new_errors)[:3]:
+                    truncated = error[:250] + "..." if len(error) > 250 else error
+                    print(f"  - {truncated}")
+            return False, new_errors
+        else:
+            # All errors existed in original
+            if verbose:
+                print(
+                    f"PASSED - No new errors (original had {len(current_errors)} errors)"
+                )
+            return True, set()
+
+    def validate_against_xsd(self):
+        """Validate XML files against XSD schemas, showing only new errors compared to original."""
+        new_errors = []
+        original_error_count = 0
+        valid_count = 0
+        skipped_count = 0
+
+        for xml_file in self.xml_files:
+            relative_path = str(xml_file.relative_to(self.unpacked_dir))
+            is_valid, new_file_errors = self.validate_file_against_xsd(
+                xml_file, verbose=False
+            )
+
+            if is_valid is None:
+                skipped_count += 1
+                continue
+            elif is_valid and not new_file_errors:
+                valid_count += 1
+                continue
+            elif is_valid:
+                # Had errors but all existed in original
+                original_error_count += 1
+                valid_count += 1
+                continue
+
+            # Has new errors
+            new_errors.append(f"  {relative_path}: {len(new_file_errors)} new error(s)")
+            for error in list(new_file_errors)[:3]:  # Show first 3 errors
+                new_errors.append(
+                    f"    - {error[:250]}..." if len(error) > 250 else f"    - {error}"
+                )
+
+        # Print summary
+        if self.verbose:
+            print(f"Validated {len(self.xml_files)} files:")
+            print(f"  - Valid: {valid_count}")
+            print(f"  - Skipped (no schema): {skipped_count}")
+            if original_error_count:
+                print(f"  - With original errors (ignored): {original_error_count}")
+            print(
+                f"  - With NEW errors: {len(new_errors) > 0 and len([e for e in new_errors if not e.startswith('    ')]) or 0}"
+            )
+
+        if new_errors:
+            print("\nFAILED - Found NEW validation errors:")
+            for error in new_errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("\nPASSED - No new XSD validation errors introduced")
+            return True
+
+    def _get_schema_path(self, xml_file):
+        """Determine the appropriate schema path for an XML file."""
+        # Check exact filename match
+        if xml_file.name in self.SCHEMA_MAPPINGS:
+            return self.schemas_dir / self.SCHEMA_MAPPINGS[xml_file.name]
+
+        # Check .rels files
+        if xml_file.suffix == ".rels":
+            return self.schemas_dir / self.SCHEMA_MAPPINGS[".rels"]
+
+        # Check chart files
+        if "charts/" in str(xml_file) and xml_file.name.startswith("chart"):
+            return self.schemas_dir / self.SCHEMA_MAPPINGS["chart"]
+
+        # Check theme files
+        if "theme/" in str(xml_file) and xml_file.name.startswith("theme"):
+            return self.schemas_dir / self.SCHEMA_MAPPINGS["theme"]
+
+        # Check if file is in a main content folder and use appropriate schema
+        if xml_file.parent.name in self.MAIN_CONTENT_FOLDERS:
+            return self.schemas_dir / self.SCHEMA_MAPPINGS[xml_file.parent.name]
+
+        return None
+
+    def _clean_ignorable_namespaces(self, xml_doc):
+        """Remove attributes and elements not in allowed namespaces."""
+        # Create a clean copy
+        xml_string = lxml.etree.tostring(xml_doc, encoding="unicode")
+        xml_copy = lxml.etree.fromstring(xml_string)
+
+        # Remove attributes not in allowed namespaces
+        for elem in xml_copy.iter():
+            attrs_to_remove = []
+
+            for attr in elem.attrib:
+                # Check if attribute is from a namespace other than allowed ones
+                if "{" in attr:
+                    ns = attr.split("}")[0][1:]
+                    if ns not in self.OOXML_NAMESPACES:
+                        attrs_to_remove.append(attr)
+
+            # Remove collected attributes
+            for attr in attrs_to_remove:
+                del elem.attrib[attr]
+
+        # Remove elements not in allowed namespaces
+        self._remove_ignorable_elements(xml_copy)
+
+        return lxml.etree.ElementTree(xml_copy)
+
+    def _remove_ignorable_elements(self, root):
+        """Recursively remove all elements not in allowed namespaces."""
+        elements_to_remove = []
+
+        # Find elements to remove
+        for elem in list(root):
+            # Skip non-element nodes (comments, processing instructions, etc.)
+            if not hasattr(elem, "tag") or callable(elem.tag):
+                continue
+
+            tag_str = str(elem.tag)
+            if tag_str.startswith("{"):
+                ns = tag_str.split("}")[0][1:]
+                if ns not in self.OOXML_NAMESPACES:
+                    elements_to_remove.append(elem)
+                    continue
+
+            # Recursively clean child elements
+            self._remove_ignorable_elements(elem)
+
+        # Remove collected elements
+        for elem in elements_to_remove:
+            root.remove(elem)
+
+    def _preprocess_for_mc_ignorable(self, xml_doc):
+        """Preprocess XML to handle mc:Ignorable attribute properly."""
+        # Remove mc:Ignorable attributes before validation
+        root = xml_doc.getroot()
+
+        # Remove mc:Ignorable attribute from root
+        if f"{{{self.MC_NAMESPACE}}}Ignorable" in root.attrib:
+            del root.attrib[f"{{{self.MC_NAMESPACE}}}Ignorable"]
+
+        return xml_doc
+
+    def _validate_single_file_xsd(self, xml_file, base_path):
+        """Validate a single XML file against XSD schema. Returns (is_valid, errors_set)."""
+        schema_path = self._get_schema_path(xml_file)
+        if not schema_path:
+            return None, None  # Skip file
+
+        try:
+            # Load schema
+            with open(schema_path, "rb") as xsd_file:
+                parser = lxml.etree.XMLParser()
+                xsd_doc = lxml.etree.parse(
+                    xsd_file, parser=parser, base_url=str(schema_path)
+                )
+                schema = lxml.etree.XMLSchema(xsd_doc)
+
+            # Load and preprocess XML
+            with open(xml_file, "r") as f:
+                xml_doc = lxml.etree.parse(f)
+
+            xml_doc, _ = self._remove_template_tags_from_text_nodes(xml_doc)
+            xml_doc = self._preprocess_for_mc_ignorable(xml_doc)
+
+            # Clean ignorable namespaces if needed
+            relative_path = xml_file.relative_to(base_path)
+            if (
+                relative_path.parts
+                and relative_path.parts[0] in self.MAIN_CONTENT_FOLDERS
+            ):
+                xml_doc = self._clean_ignorable_namespaces(xml_doc)
+
+            # Validate
+            if schema.validate(xml_doc):
+                return True, set()
+            else:
+                errors = set()
+                for error in schema.error_log:
+                    # Store normalized error message (without line numbers for comparison)
+                    errors.add(error.message)
+                return False, errors
+
+        except Exception as e:
+            return False, {str(e)}
+
+    def _get_original_file_errors(self, xml_file):
+        """Get XSD validation errors from a single file in the original document.
+
+        Args:
+            xml_file: Path to the XML file in unpacked_dir to check
+
+        Returns:
+            set: Set of error messages from the original file
+        """
+        import tempfile
+        import zipfile
+
+        # Resolve both paths to handle symlinks (e.g., /var vs /private/var on macOS)
+        xml_file = Path(xml_file).resolve()
+        unpacked_dir = self.unpacked_dir.resolve()
+        relative_path = xml_file.relative_to(unpacked_dir)
+
+        with tempfile.TemporaryDirectory() as temp_dir:
+            temp_path = Path(temp_dir)
+
+            # Extract original file
+            with zipfile.ZipFile(self.original_file, "r") as zip_ref:
+                zip_ref.extractall(temp_path)
+
+            # Find corresponding file in original
+            original_xml_file = temp_path / relative_path
+
+            if not original_xml_file.exists():
+                # File didn't exist in original, so no original errors
+                return set()
+
+            # Validate the specific file in original
+            is_valid, errors = self._validate_single_file_xsd(
+                original_xml_file, temp_path
+            )
+            return errors if errors else set()
+
+    def _remove_template_tags_from_text_nodes(self, xml_doc):
+        """Remove template tags from XML text nodes and collect warnings.
+
+        Template tags follow the pattern {{ ... }} and are used as placeholders
+        for content replacement. They should be removed from text content before
+        XSD validation while preserving XML structure.
+
+        Returns:
+            tuple: (cleaned_xml_doc, warnings_list)
+        """
+        warnings = []
+        template_pattern = re.compile(r"\{\{[^}]*\}\}")
+
+        # Create a copy of the document to avoid modifying the original
+        xml_string = lxml.etree.tostring(xml_doc, encoding="unicode")
+        xml_copy = lxml.etree.fromstring(xml_string)
+
+        def process_text_content(text, content_type):
+            if not text:
+                return text
+            matches = list(template_pattern.finditer(text))
+            if matches:
+                for match in matches:
+                    warnings.append(
+                        f"Found template tag in {content_type}: {match.group()}"
+                    )
+                return template_pattern.sub("", text)
+            return text
+
+        # Process all text nodes in the document
+        for elem in xml_copy.iter():
+            # Skip processing if this is a w:t element
+            if not hasattr(elem, "tag") or callable(elem.tag):
+                continue
+            tag_str = str(elem.tag)
+            if tag_str.endswith("}t") or tag_str == "t":
+                continue
+
+            elem.text = process_text_content(elem.text, "text content")
+            elem.tail = process_text_content(elem.tail, "tail content")
+
+        return lxml.etree.ElementTree(xml_copy), warnings
+
+
+if __name__ == "__main__":
+    raise RuntimeError("This module should not be run directly.")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validation/docx.py

@@ -0,0 +1,274 @@
+"""
+Validator for Word document XML files against XSD schemas.
+"""
+
+import re
+import tempfile
+import zipfile
+
+import lxml.etree
+
+from .base import BaseSchemaValidator
+
+
+class DOCXSchemaValidator(BaseSchemaValidator):
+    """Validator for Word document XML files against XSD schemas."""
+
+    # Word-specific namespace
+    WORD_2006_NAMESPACE = "http://schemas.openxmlformats.org/wordprocessingml/2006/main"
+
+    # Word-specific element to relationship type mappings
+    # Start with empty mapping - add specific cases as we discover them
+    ELEMENT_RELATIONSHIP_TYPES = {}
+
+    def validate(self):
+        """Run all validation checks and return True if all pass."""
+        # Test 0: XML well-formedness
+        if not self.validate_xml():
+            return False
+
+        # Test 1: Namespace declarations
+        all_valid = True
+        if not self.validate_namespaces():
+            all_valid = False
+
+        # Test 2: Unique IDs
+        if not self.validate_unique_ids():
+            all_valid = False
+
+        # Test 3: Relationship and file reference validation
+        if not self.validate_file_references():
+            all_valid = False
+
+        # Test 4: Content type declarations
+        if not self.validate_content_types():
+            all_valid = False
+
+        # Test 5: XSD schema validation
+        if not self.validate_against_xsd():
+            all_valid = False
+
+        # Test 6: Whitespace preservation
+        if not self.validate_whitespace_preservation():
+            all_valid = False
+
+        # Test 7: Deletion validation
+        if not self.validate_deletions():
+            all_valid = False
+
+        # Test 8: Insertion validation
+        if not self.validate_insertions():
+            all_valid = False
+
+        # Test 9: Relationship ID reference validation
+        if not self.validate_all_relationship_ids():
+            all_valid = False
+
+        # Count and compare paragraphs
+        self.compare_paragraph_counts()
+
+        return all_valid
+
+    def validate_whitespace_preservation(self):
+        """
+        Validate that w:t elements with whitespace have xml:space='preserve'.
+        """
+        errors = []
+
+        for xml_file in self.xml_files:
+            # Only check document.xml files
+            if xml_file.name != "document.xml":
+                continue
+
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+
+                # Find all w:t elements
+                for elem in root.iter(f"{{{self.WORD_2006_NAMESPACE}}}t"):
+                    if elem.text:
+                        text = elem.text
+                        # Check if text starts or ends with whitespace
+                        if re.match(r"^\s.*", text) or re.match(r".*\s$", text):
+                            # Check if xml:space="preserve" attribute exists
+                            xml_space_attr = f"{{{self.XML_NAMESPACE}}}space"
+                            if (
+                                xml_space_attr not in elem.attrib
+                                or elem.attrib[xml_space_attr] != "preserve"
+                            ):
+                                # Show a preview of the text
+                                text_preview = (
+                                    repr(text)[:50] + "..."
+                                    if len(repr(text)) > 50
+                                    else repr(text)
+                                )
+                                errors.append(
+                                    f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                                    f"Line {elem.sourceline}: w:t element with whitespace missing xml:space='preserve': {text_preview}"
+                                )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} whitespace preservation violations:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All whitespace is properly preserved")
+            return True
+
+    def validate_deletions(self):
+        """
+        Validate that w:t elements are not within w:del elements.
+        For some reason, XSD validation does not catch this, so we do it manually.
+        """
+        errors = []
+
+        for xml_file in self.xml_files:
+            # Only check document.xml files
+            if xml_file.name != "document.xml":
+                continue
+
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+
+                # Find all w:t elements that are descendants of w:del elements
+                namespaces = {"w": self.WORD_2006_NAMESPACE}
+                xpath_expression = ".//w:del//w:t"
+                problematic_t_elements = root.xpath(
+                    xpath_expression, namespaces=namespaces
+                )
+                for t_elem in problematic_t_elements:
+                    if t_elem.text:
+                        # Show a preview of the text
+                        text_preview = (
+                            repr(t_elem.text)[:50] + "..."
+                            if len(repr(t_elem.text)) > 50
+                            else repr(t_elem.text)
+                        )
+                        errors.append(
+                            f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                            f"Line {t_elem.sourceline}: <w:t> found within <w:del>: {text_preview}"
+                        )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} deletion validation violations:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - No w:t elements found within w:del elements")
+            return True
+
+    def count_paragraphs_in_unpacked(self):
+        """Count the number of paragraphs in the unpacked document."""
+        count = 0
+
+        for xml_file in self.xml_files:
+            # Only check document.xml files
+            if xml_file.name != "document.xml":
+                continue
+
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+                # Count all w:p elements
+                paragraphs = root.findall(f".//{{{self.WORD_2006_NAMESPACE}}}p")
+                count = len(paragraphs)
+            except Exception as e:
+                print(f"Error counting paragraphs in unpacked document: {e}")
+
+        return count
+
+    def count_paragraphs_in_original(self):
+        """Count the number of paragraphs in the original docx file."""
+        count = 0
+
+        try:
+            # Create temporary directory to unpack original
+            with tempfile.TemporaryDirectory() as temp_dir:
+                # Unpack original docx
+                with zipfile.ZipFile(self.original_file, "r") as zip_ref:
+                    zip_ref.extractall(temp_dir)
+
+                # Parse document.xml
+                doc_xml_path = temp_dir + "/word/document.xml"
+                root = lxml.etree.parse(doc_xml_path).getroot()
+
+                # Count all w:p elements
+                paragraphs = root.findall(f".//{{{self.WORD_2006_NAMESPACE}}}p")
+                count = len(paragraphs)
+
+        except Exception as e:
+            print(f"Error counting paragraphs in original document: {e}")
+
+        return count
+
+    def validate_insertions(self):
+        """
+        Validate that w:delText elements are not within w:ins elements.
+        w:delText is only allowed in w:ins if nested within a w:del.
+        """
+        errors = []
+
+        for xml_file in self.xml_files:
+            if xml_file.name != "document.xml":
+                continue
+
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+                namespaces = {"w": self.WORD_2006_NAMESPACE}
+
+                # Find w:delText in w:ins that are NOT within w:del
+                invalid_elements = root.xpath(
+                    ".//w:ins//w:delText[not(ancestor::w:del)]",
+                    namespaces=namespaces
+                )
+
+                for elem in invalid_elements:
+                    text_preview = (
+                        repr(elem.text or "")[:50] + "..."
+                        if len(repr(elem.text or "")) > 50
+                        else repr(elem.text or "")
+                    )
+                    errors.append(
+                        f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                        f"Line {elem.sourceline}: <w:delText> within <w:ins>: {text_preview}"
+                    )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} insertion validation violations:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - No w:delText elements within w:ins elements")
+            return True
+
+    def compare_paragraph_counts(self):
+        """Compare paragraph counts between original and new document."""
+        original_count = self.count_paragraphs_in_original()
+        new_count = self.count_paragraphs_in_unpacked()
+
+        diff = new_count - original_count
+        diff_str = f"+{diff}" if diff > 0 else str(diff)
+        print(f"\nParagraphs: {original_count} → {new_count} ({diff_str})")
+
+
+if __name__ == "__main__":
+    raise RuntimeError("This module should not be run directly.")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validation/pptx.py

@@ -0,0 +1,315 @@
+"""
+Validator for PowerPoint presentation XML files against XSD schemas.
+"""
+
+import re
+
+from .base import BaseSchemaValidator
+
+
+class PPTXSchemaValidator(BaseSchemaValidator):
+    """Validator for PowerPoint presentation XML files against XSD schemas."""
+
+    # PowerPoint presentation namespace
+    PRESENTATIONML_NAMESPACE = (
+        "http://schemas.openxmlformats.org/presentationml/2006/main"
+    )
+
+    # PowerPoint-specific element to relationship type mappings
+    ELEMENT_RELATIONSHIP_TYPES = {
+        "sldid": "slide",
+        "sldmasterid": "slidemaster",
+        "notesmasterid": "notesmaster",
+        "sldlayoutid": "slidelayout",
+        "themeid": "theme",
+        "tablestyleid": "tablestyles",
+    }
+
+    def validate(self):
+        """Run all validation checks and return True if all pass."""
+        # Test 0: XML well-formedness
+        if not self.validate_xml():
+            return False
+
+        # Test 1: Namespace declarations
+        all_valid = True
+        if not self.validate_namespaces():
+            all_valid = False
+
+        # Test 2: Unique IDs
+        if not self.validate_unique_ids():
+            all_valid = False
+
+        # Test 3: UUID ID validation
+        if not self.validate_uuid_ids():
+            all_valid = False
+
+        # Test 4: Relationship and file reference validation
+        if not self.validate_file_references():
+            all_valid = False
+
+        # Test 5: Slide layout ID validation
+        if not self.validate_slide_layout_ids():
+            all_valid = False
+
+        # Test 6: Content type declarations
+        if not self.validate_content_types():
+            all_valid = False
+
+        # Test 7: XSD schema validation
+        if not self.validate_against_xsd():
+            all_valid = False
+
+        # Test 8: Notes slide reference validation
+        if not self.validate_notes_slide_references():
+            all_valid = False
+
+        # Test 9: Relationship ID reference validation
+        if not self.validate_all_relationship_ids():
+            all_valid = False
+
+        # Test 10: Duplicate slide layout references validation
+        if not self.validate_no_duplicate_slide_layouts():
+            all_valid = False
+
+        return all_valid
+
+    def validate_uuid_ids(self):
+        """Validate that ID attributes that look like UUIDs contain only hex values."""
+        import lxml.etree
+
+        errors = []
+        # UUID pattern: 8-4-4-4-12 hex digits with optional braces/hyphens
+        uuid_pattern = re.compile(
+            r"^[\{\(]?[0-9A-Fa-f]{8}-?[0-9A-Fa-f]{4}-?[0-9A-Fa-f]{4}-?[0-9A-Fa-f]{4}-?[0-9A-Fa-f]{12}[\}\)]?$"
+        )
+
+        for xml_file in self.xml_files:
+            try:
+                root = lxml.etree.parse(str(xml_file)).getroot()
+
+                # Check all elements for ID attributes
+                for elem in root.iter():
+                    for attr, value in elem.attrib.items():
+                        # Check if this is an ID attribute
+                        attr_name = attr.split("}")[-1].lower()
+                        if attr_name == "id" or attr_name.endswith("id"):
+                            # Check if value looks like a UUID (has the right length and pattern structure)
+                            if self._looks_like_uuid(value):
+                                # Validate that it contains only hex characters in the right positions
+                                if not uuid_pattern.match(value):
+                                    errors.append(
+                                        f"  {xml_file.relative_to(self.unpacked_dir)}: "
+                                        f"Line {elem.sourceline}: ID '{value}' appears to be a UUID but contains invalid hex characters"
+                                    )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {xml_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} UUID ID validation errors:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All UUID-like IDs contain valid hex values")
+            return True
+
+    def _looks_like_uuid(self, value):
+        """Check if a value has the general structure of a UUID."""
+        # Remove common UUID delimiters
+        clean_value = value.strip("{}()").replace("-", "")
+        # Check if it's 32 hex-like characters (could include invalid hex chars)
+        return len(clean_value) == 32 and all(c.isalnum() for c in clean_value)
+
+    def validate_slide_layout_ids(self):
+        """Validate that sldLayoutId elements in slide masters reference valid slide layouts."""
+        import lxml.etree
+
+        errors = []
+
+        # Find all slide master files
+        slide_masters = list(self.unpacked_dir.glob("ppt/slideMasters/*.xml"))
+
+        if not slide_masters:
+            if self.verbose:
+                print("PASSED - No slide masters found")
+            return True
+
+        for slide_master in slide_masters:
+            try:
+                # Parse the slide master file
+                root = lxml.etree.parse(str(slide_master)).getroot()
+
+                # Find the corresponding _rels file for this slide master
+                rels_file = slide_master.parent / "_rels" / f"{slide_master.name}.rels"
+
+                if not rels_file.exists():
+                    errors.append(
+                        f"  {slide_master.relative_to(self.unpacked_dir)}: "
+                        f"Missing relationships file: {rels_file.relative_to(self.unpacked_dir)}"
+                    )
+                    continue
+
+                # Parse the relationships file
+                rels_root = lxml.etree.parse(str(rels_file)).getroot()
+
+                # Build a set of valid relationship IDs that point to slide layouts
+                valid_layout_rids = set()
+                for rel in rels_root.findall(
+                    f".//{{{self.PACKAGE_RELATIONSHIPS_NAMESPACE}}}Relationship"
+                ):
+                    rel_type = rel.get("Type", "")
+                    if "slideLayout" in rel_type:
+                        valid_layout_rids.add(rel.get("Id"))
+
+                # Find all sldLayoutId elements in the slide master
+                for sld_layout_id in root.findall(
+                    f".//{{{self.PRESENTATIONML_NAMESPACE}}}sldLayoutId"
+                ):
+                    r_id = sld_layout_id.get(
+                        f"{{{self.OFFICE_RELATIONSHIPS_NAMESPACE}}}id"
+                    )
+                    layout_id = sld_layout_id.get("id")
+
+                    if r_id and r_id not in valid_layout_rids:
+                        errors.append(
+                            f"  {slide_master.relative_to(self.unpacked_dir)}: "
+                            f"Line {sld_layout_id.sourceline}: sldLayoutId with id='{layout_id}' "
+                            f"references r:id='{r_id}' which is not found in slide layout relationships"
+                        )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {slide_master.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print(f"FAILED - Found {len(errors)} slide layout ID validation errors:")
+            for error in errors:
+                print(error)
+            print(
+                "Remove invalid references or add missing slide layouts to the relationships file."
+            )
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All slide layout IDs reference valid slide layouts")
+            return True
+
+    def validate_no_duplicate_slide_layouts(self):
+        """Validate that each slide has exactly one slideLayout reference."""
+        import lxml.etree
+
+        errors = []
+        slide_rels_files = list(self.unpacked_dir.glob("ppt/slides/_rels/*.xml.rels"))
+
+        for rels_file in slide_rels_files:
+            try:
+                root = lxml.etree.parse(str(rels_file)).getroot()
+
+                # Find all slideLayout relationships
+                layout_rels = [
+                    rel
+                    for rel in root.findall(
+                        f".//{{{self.PACKAGE_RELATIONSHIPS_NAMESPACE}}}Relationship"
+                    )
+                    if "slideLayout" in rel.get("Type", "")
+                ]
+
+                if len(layout_rels) > 1:
+                    errors.append(
+                        f"  {rels_file.relative_to(self.unpacked_dir)}: has {len(layout_rels)} slideLayout references"
+                    )
+
+            except Exception as e:
+                errors.append(
+                    f"  {rels_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        if errors:
+            print("FAILED - Found slides with duplicate slideLayout references:")
+            for error in errors:
+                print(error)
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All slides have exactly one slideLayout reference")
+            return True
+
+    def validate_notes_slide_references(self):
+        """Validate that each notesSlide file is referenced by only one slide."""
+        import lxml.etree
+
+        errors = []
+        notes_slide_references = {}  # Track which slides reference each notesSlide
+
+        # Find all slide relationship files
+        slide_rels_files = list(self.unpacked_dir.glob("ppt/slides/_rels/*.xml.rels"))
+
+        if not slide_rels_files:
+            if self.verbose:
+                print("PASSED - No slide relationship files found")
+            return True
+
+        for rels_file in slide_rels_files:
+            try:
+                # Parse the relationships file
+                root = lxml.etree.parse(str(rels_file)).getroot()
+
+                # Find all notesSlide relationships
+                for rel in root.findall(
+                    f".//{{{self.PACKAGE_RELATIONSHIPS_NAMESPACE}}}Relationship"
+                ):
+                    rel_type = rel.get("Type", "")
+                    if "notesSlide" in rel_type:
+                        target = rel.get("Target", "")
+                        if target:
+                            # Normalize the target path to handle relative paths
+                            normalized_target = target.replace("../", "")
+
+                            # Track which slide references this notesSlide
+                            slide_name = rels_file.stem.replace(
+                                ".xml", ""
+                            )  # e.g., "slide1"
+
+                            if normalized_target not in notes_slide_references:
+                                notes_slide_references[normalized_target] = []
+                            notes_slide_references[normalized_target].append(
+                                (slide_name, rels_file)
+                            )
+
+            except (lxml.etree.XMLSyntaxError, Exception) as e:
+                errors.append(
+                    f"  {rels_file.relative_to(self.unpacked_dir)}: Error: {e}"
+                )
+
+        # Check for duplicate references
+        for target, references in notes_slide_references.items():
+            if len(references) > 1:
+                slide_names = [ref[0] for ref in references]
+                errors.append(
+                    f"  Notes slide '{target}' is referenced by multiple slides: {', '.join(slide_names)}"
+                )
+                for slide_name, rels_file in references:
+                    errors.append(f"    - {rels_file.relative_to(self.unpacked_dir)}")
+
+        if errors:
+            print(
+                f"FAILED - Found {len([e for e in errors if not e.startswith('    ')])} notes slide reference validation errors:"
+            )
+            for error in errors:
+                print(error)
+            print("Each slide may optionally have its own slide file.")
+            return False
+        else:
+            if self.verbose:
+                print("PASSED - All notes slide references are unique")
+            return True
+
+
+if __name__ == "__main__":
+    raise RuntimeError("This module should not be run directly.")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/docx/validation/redlining.py

@@ -0,0 +1,279 @@
+"""
+Validator for tracked changes in Word documents.
+"""
+
+import subprocess
+import tempfile
+import zipfile
+from pathlib import Path
+
+
+class RedliningValidator:
+    """Validator for tracked changes in Word documents."""
+
+    def __init__(self, unpacked_dir, original_docx, verbose=False):
+        self.unpacked_dir = Path(unpacked_dir)
+        self.original_docx = Path(original_docx)
+        self.verbose = verbose
+        self.namespaces = {
+            "w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main"
+        }
+
+    def validate(self):
+        """Main validation method that returns True if valid, False otherwise."""
+        # Verify unpacked directory exists and has correct structure
+        modified_file = self.unpacked_dir / "word" / "document.xml"
+        if not modified_file.exists():
+            print(f"FAILED - Modified document.xml not found at {modified_file}")
+            return False
+
+        # First, check if there are any tracked changes by Claude to validate
+        try:
+            import xml.etree.ElementTree as ET
+
+            tree = ET.parse(modified_file)
+            root = tree.getroot()
+
+            # Check for w:del or w:ins tags authored by Claude
+            del_elements = root.findall(".//w:del", self.namespaces)
+            ins_elements = root.findall(".//w:ins", self.namespaces)
+
+            # Filter to only include changes by Claude
+            claude_del_elements = [
+                elem
+                for elem in del_elements
+                if elem.get(f"{{{self.namespaces['w']}}}author") == "Claude"
+            ]
+            claude_ins_elements = [
+                elem
+                for elem in ins_elements
+                if elem.get(f"{{{self.namespaces['w']}}}author") == "Claude"
+            ]
+
+            # Redlining validation is only needed if tracked changes by Claude have been used.
+            if not claude_del_elements and not claude_ins_elements:
+                if self.verbose:
+                    print("PASSED - No tracked changes by Claude found.")
+                return True
+
+        except Exception:
+            # If we can't parse the XML, continue with full validation
+            pass
+
+        # Create temporary directory for unpacking original docx
+        with tempfile.TemporaryDirectory() as temp_dir:
+            temp_path = Path(temp_dir)
+
+            # Unpack original docx
+            try:
+                with zipfile.ZipFile(self.original_docx, "r") as zip_ref:
+                    zip_ref.extractall(temp_path)
+            except Exception as e:
+                print(f"FAILED - Error unpacking original docx: {e}")
+                return False
+
+            original_file = temp_path / "word" / "document.xml"
+            if not original_file.exists():
+                print(
+                    f"FAILED - Original document.xml not found in {self.original_docx}"
+                )
+                return False
+
+            # Parse both XML files using xml.etree.ElementTree for redlining validation
+            try:
+                import xml.etree.ElementTree as ET
+
+                modified_tree = ET.parse(modified_file)
+                modified_root = modified_tree.getroot()
+                original_tree = ET.parse(original_file)
+                original_root = original_tree.getroot()
+            except ET.ParseError as e:
+                print(f"FAILED - Error parsing XML files: {e}")
+                return False
+
+            # Remove Claude's tracked changes from both documents
+            self._remove_claude_tracked_changes(original_root)
+            self._remove_claude_tracked_changes(modified_root)
+
+            # Extract and compare text content
+            modified_text = self._extract_text_content(modified_root)
+            original_text = self._extract_text_content(original_root)
+
+            if modified_text != original_text:
+                # Show detailed character-level differences for each paragraph
+                error_message = self._generate_detailed_diff(
+                    original_text, modified_text
+                )
+                print(error_message)
+                return False
+
+            if self.verbose:
+                print("PASSED - All changes by Claude are properly tracked")
+            return True
+
+    def _generate_detailed_diff(self, original_text, modified_text):
+        """Generate detailed word-level differences using git word diff."""
+        error_parts = [
+            "FAILED - Document text doesn't match after removing Claude's tracked changes",
+            "",
+            "Likely causes:",
+            "  1. Modified text inside another author's <w:ins> or <w:del> tags",
+            "  2. Made edits without proper tracked changes",
+            "  3. Didn't nest <w:del> inside <w:ins> when deleting another's insertion",
+            "",
+            "For pre-redlined documents, use correct patterns:",
+            "  - To reject another's INSERTION: Nest <w:del> inside their <w:ins>",
+            "  - To restore another's DELETION: Add new <w:ins> AFTER their <w:del>",
+            "",
+        ]
+
+        # Show git word diff
+        git_diff = self._get_git_word_diff(original_text, modified_text)
+        if git_diff:
+            error_parts.extend(["Differences:", "============", git_diff])
+        else:
+            error_parts.append("Unable to generate word diff (git not available)")
+
+        return "\n".join(error_parts)
+
+    def _get_git_word_diff(self, original_text, modified_text):
+        """Generate word diff using git with character-level precision."""
+        try:
+            with tempfile.TemporaryDirectory() as temp_dir:
+                temp_path = Path(temp_dir)
+
+                # Create two files
+                original_file = temp_path / "original.txt"
+                modified_file = temp_path / "modified.txt"
+
+                original_file.write_text(original_text, encoding="utf-8")
+                modified_file.write_text(modified_text, encoding="utf-8")
+
+                # Try character-level diff first for precise differences
+                result = subprocess.run(
+                    [
+                        "git",
+                        "diff",
+                        "--word-diff=plain",
+                        "--word-diff-regex=.",  # Character-by-character diff
+                        "-U0",  # Zero lines of context - show only changed lines
+                        "--no-index",
+                        str(original_file),
+                        str(modified_file),
+                    ],
+                    capture_output=True,
+                    text=True,
+                )
+
+                if result.stdout.strip():
+                    # Clean up the output - remove git diff header lines
+                    lines = result.stdout.split("\n")
+                    # Skip the header lines (diff --git, index, +++, ---, @@)
+                    content_lines = []
+                    in_content = False
+                    for line in lines:
+                        if line.startswith("@@"):
+                            in_content = True
+                            continue
+                        if in_content and line.strip():
+                            content_lines.append(line)
+
+                    if content_lines:
+                        return "\n".join(content_lines)
+
+                # Fallback to word-level diff if character-level is too verbose
+                result = subprocess.run(
+                    [
+                        "git",
+                        "diff",
+                        "--word-diff=plain",
+                        "-U0",  # Zero lines of context
+                        "--no-index",
+                        str(original_file),
+                        str(modified_file),
+                    ],
+                    capture_output=True,
+                    text=True,
+                )
+
+                if result.stdout.strip():
+                    lines = result.stdout.split("\n")
+                    content_lines = []
+                    in_content = False
+                    for line in lines:
+                        if line.startswith("@@"):
+                            in_content = True
+                            continue
+                        if in_content and line.strip():
+                            content_lines.append(line)
+                    return "\n".join(content_lines)
+
+        except (subprocess.CalledProcessError, FileNotFoundError, Exception):
+            # Git not available or other error, return None to use fallback
+            pass
+
+        return None
+
+    def _remove_claude_tracked_changes(self, root):
+        """Remove tracked changes authored by Claude from the XML root."""
+        ins_tag = f"{{{self.namespaces['w']}}}ins"
+        del_tag = f"{{{self.namespaces['w']}}}del"
+        author_attr = f"{{{self.namespaces['w']}}}author"
+
+        # Remove w:ins elements
+        for parent in root.iter():
+            to_remove = []
+            for child in parent:
+                if child.tag == ins_tag and child.get(author_attr) == "Claude":
+                    to_remove.append(child)
+            for elem in to_remove:
+                parent.remove(elem)
+
+        # Unwrap content in w:del elements where author is "Claude"
+        deltext_tag = f"{{{self.namespaces['w']}}}delText"
+        t_tag = f"{{{self.namespaces['w']}}}t"
+
+        for parent in root.iter():
+            to_process = []
+            for child in parent:
+                if child.tag == del_tag and child.get(author_attr) == "Claude":
+                    to_process.append((child, list(parent).index(child)))
+
+            # Process in reverse order to maintain indices
+            for del_elem, del_index in reversed(to_process):
+                # Convert w:delText to w:t before moving
+                for elem in del_elem.iter():
+                    if elem.tag == deltext_tag:
+                        elem.tag = t_tag
+
+                # Move all children of w:del to its parent before removing w:del
+                for child in reversed(list(del_elem)):
+                    parent.insert(del_index, child)
+                parent.remove(del_elem)
+
+    def _extract_text_content(self, root):
+        """Extract text content from Word XML, preserving paragraph structure.
+
+        Empty paragraphs are skipped to avoid false positives when tracked
+        insertions add only structural elements without text content.
+        """
+        p_tag = f"{{{self.namespaces['w']}}}p"
+        t_tag = f"{{{self.namespaces['w']}}}t"
+
+        paragraphs = []
+        for p_elem in root.findall(f".//{p_tag}"):
+            # Get all text elements within this paragraph
+            text_parts = []
+            for t_elem in p_elem.findall(f".//{t_tag}"):
+                if t_elem.text:
+                    text_parts.append(t_elem.text)
+            paragraph_text = "".join(text_parts)
+            # Skip empty paragraphs - they don't affect content validation
+            if paragraph_text:
+                paragraphs.append(paragraph_text)
+
+        return "\n".join(paragraphs)
+
+
+if __name__ == "__main__":
+    raise RuntimeError("This module should not be run directly.")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/check_bounding_boxes.py

@@ -0,0 +1,70 @@
+from dataclasses import dataclass
+import json
+import sys
+
+
+# Script to check that the `fields.json` file that Claude creates when analyzing PDFs
+# does not have overlapping bounding boxes. See forms.md.
+
+
+@dataclass
+class RectAndField:
+    rect: list[float]
+    rect_type: str
+    field: dict
+
+
+# Returns a list of messages that are printed to stdout for Claude to read.
+def get_bounding_box_messages(fields_json_stream) -> list[str]:
+    messages = []
+    fields = json.load(fields_json_stream)
+    messages.append(f"Read {len(fields['form_fields'])} fields")
+
+    def rects_intersect(r1, r2):
+        disjoint_horizontal = r1[0] >= r2[2] or r1[2] <= r2[0]
+        disjoint_vertical = r1[1] >= r2[3] or r1[3] <= r2[1]
+        return not (disjoint_horizontal or disjoint_vertical)
+
+    rects_and_fields = []
+    for f in fields["form_fields"]:
+        rects_and_fields.append(RectAndField(f["label_bounding_box"], "label", f))
+        rects_and_fields.append(RectAndField(f["entry_bounding_box"], "entry", f))
+
+    has_error = False
+    for i, ri in enumerate(rects_and_fields):
+        # This is O(N^2); we can optimize if it becomes a problem.
+        for j in range(i + 1, len(rects_and_fields)):
+            rj = rects_and_fields[j]
+            if ri.field["page_number"] == rj.field["page_number"] and rects_intersect(ri.rect, rj.rect):
+                has_error = True
+                if ri.field is rj.field:
+                    messages.append(f"FAILURE: intersection between label and entry bounding boxes for `{ri.field['description']}` ({ri.rect}, {rj.rect})")
+                else:
+                    messages.append(f"FAILURE: intersection between {ri.rect_type} bounding box for `{ri.field['description']}` ({ri.rect}) and {rj.rect_type} bounding box for `{rj.field['description']}` ({rj.rect})")
+                if len(messages) >= 20:
+                    messages.append("Aborting further checks; fix bounding boxes and try again")
+                    return messages
+        if ri.rect_type == "entry":
+            if "entry_text" in ri.field:
+                font_size = ri.field["entry_text"].get("font_size", 14)
+                entry_height = ri.rect[3] - ri.rect[1]
+                if entry_height < font_size:
+                    has_error = True
+                    messages.append(f"FAILURE: entry bounding box height ({entry_height}) for `{ri.field['description']}` is too short for the text content (font size: {font_size}). Increase the box height or decrease the font size.")
+                    if len(messages) >= 20:
+                        messages.append("Aborting further checks; fix bounding boxes and try again")
+                        return messages
+
+    if not has_error:
+        messages.append("SUCCESS: All bounding boxes are valid")
+    return messages
+
+if __name__ == "__main__":
+    if len(sys.argv) != 2:
+        print("Usage: check_bounding_boxes.py [fields.json]")
+        sys.exit(1)
+    # Input file should be in the `fields.json` format described in forms.md.
+    with open(sys.argv[1]) as f:
+        messages = get_bounding_box_messages(f)
+    for msg in messages:
+        print(msg)

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/check_bounding_boxes_test.py

@@ -0,0 +1,226 @@
+import unittest
+import json
+import io
+from check_bounding_boxes import get_bounding_box_messages
+
+
+# Currently this is not run automatically in CI; it's just for documentation and manual checking.
+class TestGetBoundingBoxMessages(unittest.TestCase):
+    
+    def create_json_stream(self, data):
+        """Helper to create a JSON stream from data"""
+        return io.StringIO(json.dumps(data))
+    
+    def test_no_intersections(self):
+        """Test case with no bounding box intersections"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 30]
+                },
+                {
+                    "description": "Email",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 40, 50, 60],
+                    "entry_bounding_box": [60, 40, 150, 60]
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("SUCCESS" in msg for msg in messages))
+        self.assertFalse(any("FAILURE" in msg for msg in messages))
+    
+    def test_label_entry_intersection_same_field(self):
+        """Test intersection between label and entry of the same field"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 60, 30],
+                    "entry_bounding_box": [50, 10, 150, 30]  # Overlaps with label
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("FAILURE" in msg and "intersection" in msg for msg in messages))
+        self.assertFalse(any("SUCCESS" in msg for msg in messages))
+    
+    def test_intersection_between_different_fields(self):
+        """Test intersection between bounding boxes of different fields"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 30]
+                },
+                {
+                    "description": "Email",
+                    "page_number": 1,
+                    "label_bounding_box": [40, 20, 80, 40],  # Overlaps with Name's boxes
+                    "entry_bounding_box": [160, 10, 250, 30]
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("FAILURE" in msg and "intersection" in msg for msg in messages))
+        self.assertFalse(any("SUCCESS" in msg for msg in messages))
+    
+    def test_different_pages_no_intersection(self):
+        """Test that boxes on different pages don't count as intersecting"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 30]
+                },
+                {
+                    "description": "Email",
+                    "page_number": 2,
+                    "label_bounding_box": [10, 10, 50, 30],  # Same coordinates but different page
+                    "entry_bounding_box": [60, 10, 150, 30]
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("SUCCESS" in msg for msg in messages))
+        self.assertFalse(any("FAILURE" in msg for msg in messages))
+    
+    def test_entry_height_too_small(self):
+        """Test that entry box height is checked against font size"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 20],  # Height is 10
+                    "entry_text": {
+                        "font_size": 14  # Font size larger than height
+                    }
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("FAILURE" in msg and "height" in msg for msg in messages))
+        self.assertFalse(any("SUCCESS" in msg for msg in messages))
+    
+    def test_entry_height_adequate(self):
+        """Test that adequate entry box height passes"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 30],  # Height is 20
+                    "entry_text": {
+                        "font_size": 14  # Font size smaller than height
+                    }
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("SUCCESS" in msg for msg in messages))
+        self.assertFalse(any("FAILURE" in msg for msg in messages))
+    
+    def test_default_font_size(self):
+        """Test that default font size is used when not specified"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 20],  # Height is 10
+                    "entry_text": {}  # No font_size specified, should use default 14
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("FAILURE" in msg and "height" in msg for msg in messages))
+        self.assertFalse(any("SUCCESS" in msg for msg in messages))
+    
+    def test_no_entry_text(self):
+        """Test that missing entry_text doesn't cause height check"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [60, 10, 150, 20]  # Small height but no entry_text
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("SUCCESS" in msg for msg in messages))
+        self.assertFalse(any("FAILURE" in msg for msg in messages))
+    
+    def test_multiple_errors_limit(self):
+        """Test that error messages are limited to prevent excessive output"""
+        fields = []
+        # Create many overlapping fields
+        for i in range(25):
+            fields.append({
+                "description": f"Field{i}",
+                "page_number": 1,
+                "label_bounding_box": [10, 10, 50, 30],  # All overlap
+                "entry_bounding_box": [20, 15, 60, 35]   # All overlap
+            })
+        
+        data = {"form_fields": fields}
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        # Should abort after ~20 messages
+        self.assertTrue(any("Aborting" in msg for msg in messages))
+        # Should have some FAILURE messages but not hundreds
+        failure_count = sum(1 for msg in messages if "FAILURE" in msg)
+        self.assertGreater(failure_count, 0)
+        self.assertLess(len(messages), 30)  # Should be limited
+    
+    def test_edge_touching_boxes(self):
+        """Test that boxes touching at edges don't count as intersecting"""
+        data = {
+            "form_fields": [
+                {
+                    "description": "Name",
+                    "page_number": 1,
+                    "label_bounding_box": [10, 10, 50, 30],
+                    "entry_bounding_box": [50, 10, 150, 30]  # Touches at x=50
+                }
+            ]
+        }
+        
+        stream = self.create_json_stream(data)
+        messages = get_bounding_box_messages(stream)
+        self.assertTrue(any("SUCCESS" in msg for msg in messages))
+        self.assertFalse(any("FAILURE" in msg for msg in messages))
+    
+
+if __name__ == '__main__':
+    unittest.main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/check_fillable_fields.py

@@ -0,0 +1,12 @@
+import sys
+from pypdf import PdfReader
+
+
+# Script for Claude to run to determine whether a PDF has fillable form fields. See forms.md.
+
+
+reader = PdfReader(sys.argv[1])
+if (reader.get_fields()):
+    print("This PDF has fillable form fields")
+else:
+    print("This PDF does not have fillable form fields; you will need to visually determine where to enter data")

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/convert_pdf_to_images.py

@@ -0,0 +1,35 @@
+import os
+import sys
+
+from pdf2image import convert_from_path
+
+
+# Converts each page of a PDF to a PNG image.
+
+
+def convert(pdf_path, output_dir, max_dim=1000):
+    images = convert_from_path(pdf_path, dpi=200)
+
+    for i, image in enumerate(images):
+        # Scale image if needed to keep width/height under `max_dim`
+        width, height = image.size
+        if width > max_dim or height > max_dim:
+            scale_factor = min(max_dim / width, max_dim / height)
+            new_width = int(width * scale_factor)
+            new_height = int(height * scale_factor)
+            image = image.resize((new_width, new_height))
+        
+        image_path = os.path.join(output_dir, f"page_{i+1}.png")
+        image.save(image_path)
+        print(f"Saved page {i+1} as {image_path} (size: {image.size})")
+
+    print(f"Converted {len(images)} pages to PNG images")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 3:
+        print("Usage: convert_pdf_to_images.py [input pdf] [output directory]")
+        sys.exit(1)
+    pdf_path = sys.argv[1]
+    output_directory = sys.argv[2]
+    convert(pdf_path, output_directory)

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/create_validation_image.py

@@ -0,0 +1,41 @@
+import json
+import sys
+
+from PIL import Image, ImageDraw
+
+
+# Creates "validation" images with rectangles for the bounding box information that
+# Claude creates when determining where to add text annotations in PDFs. See forms.md.
+
+
+def create_validation_image(page_number, fields_json_path, input_path, output_path):
+    # Input file should be in the `fields.json` format described in forms.md.
+    with open(fields_json_path, 'r') as f:
+        data = json.load(f)
+
+        img = Image.open(input_path)
+        draw = ImageDraw.Draw(img)
+        num_boxes = 0
+        
+        for field in data["form_fields"]:
+            if field["page_number"] == page_number:
+                entry_box = field['entry_bounding_box']
+                label_box = field['label_bounding_box']
+                # Draw red rectangle over entry bounding box and blue rectangle over the label.
+                draw.rectangle(entry_box, outline='red', width=2)
+                draw.rectangle(label_box, outline='blue', width=2)
+                num_boxes += 2
+        
+        img.save(output_path)
+        print(f"Created validation image at {output_path} with {num_boxes} bounding boxes")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 5:
+        print("Usage: create_validation_image.py [page number] [fields.json file] [input image path] [output image path]")
+        sys.exit(1)
+    page_number = int(sys.argv[1])
+    fields_json_path = sys.argv[2]
+    input_image_path = sys.argv[3]
+    output_image_path = sys.argv[4]
+    create_validation_image(page_number, fields_json_path, input_image_path, output_image_path)

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/extract_form_field_info.py

@@ -0,0 +1,152 @@
+import json
+import sys
+
+from pypdf import PdfReader
+
+
+# Extracts data for the fillable form fields in a PDF and outputs JSON that
+# Claude uses to fill the fields. See forms.md.
+
+
+# This matches the format used by PdfReader `get_fields` and `update_page_form_field_values` methods.
+def get_full_annotation_field_id(annotation):
+    components = []
+    while annotation:
+        field_name = annotation.get('/T')
+        if field_name:
+            components.append(field_name)
+        annotation = annotation.get('/Parent')
+    return ".".join(reversed(components)) if components else None
+
+
+def make_field_dict(field, field_id):
+    field_dict = {"field_id": field_id}
+    ft = field.get('/FT')
+    if ft == "/Tx":
+        field_dict["type"] = "text"
+    elif ft == "/Btn":
+        field_dict["type"] = "checkbox"  # radio groups handled separately
+        states = field.get("/_States_", [])
+        if len(states) == 2:
+            # "/Off" seems to always be the unchecked value, as suggested by
+            # https://opensource.adobe.com/dc-acrobat-sdk-docs/standards/pdfstandards/pdf/PDF32000_2008.pdf#page=448
+            # It can be either first or second in the "/_States_" list.
+            if "/Off" in states:
+                field_dict["checked_value"] = states[0] if states[0] != "/Off" else states[1]
+                field_dict["unchecked_value"] = "/Off"
+            else:
+                print(f"Unexpected state values for checkbox `${field_id}`. Its checked and unchecked values may not be correct; if you're trying to check it, visually verify the results.")
+                field_dict["checked_value"] = states[0]
+                field_dict["unchecked_value"] = states[1]
+    elif ft == "/Ch":
+        field_dict["type"] = "choice"
+        states = field.get("/_States_", [])
+        field_dict["choice_options"] = [{
+            "value": state[0],
+            "text": state[1],
+        } for state in states]
+    else:
+        field_dict["type"] = f"unknown ({ft})"
+    return field_dict
+
+
+# Returns a list of fillable PDF fields:
+# [
+#   {
+#     "field_id": "name",
+#     "page": 1,
+#     "type": ("text", "checkbox", "radio_group", or "choice")
+#     // Per-type additional fields described in forms.md
+#   },
+# ]
+def get_field_info(reader: PdfReader):
+    fields = reader.get_fields()
+
+    field_info_by_id = {}
+    possible_radio_names = set()
+
+    for field_id, field in fields.items():
+        # Skip if this is a container field with children, except that it might be
+        # a parent group for radio button options.
+        if field.get("/Kids"):
+            if field.get("/FT") == "/Btn":
+                possible_radio_names.add(field_id)
+            continue
+        field_info_by_id[field_id] = make_field_dict(field, field_id)
+
+    # Bounding rects are stored in annotations in page objects.
+
+    # Radio button options have a separate annotation for each choice;
+    # all choices have the same field name.
+    # See https://westhealth.github.io/exploring-fillable-forms-with-pdfrw.html
+    radio_fields_by_id = {}
+
+    for page_index, page in enumerate(reader.pages):
+        annotations = page.get('/Annots', [])
+        for ann in annotations:
+            field_id = get_full_annotation_field_id(ann)
+            if field_id in field_info_by_id:
+                field_info_by_id[field_id]["page"] = page_index + 1
+                field_info_by_id[field_id]["rect"] = ann.get('/Rect')
+            elif field_id in possible_radio_names:
+                try:
+                    # ann['/AP']['/N'] should have two items. One of them is '/Off',
+                    # the other is the active value.
+                    on_values = [v for v in ann["/AP"]["/N"] if v != "/Off"]
+                except KeyError:
+                    continue
+                if len(on_values) == 1:
+                    rect = ann.get("/Rect")
+                    if field_id not in radio_fields_by_id:
+                        radio_fields_by_id[field_id] = {
+                            "field_id": field_id,
+                            "type": "radio_group",
+                            "page": page_index + 1,
+                            "radio_options": [],
+                        }
+                    # Note: at least on macOS 15.7, Preview.app doesn't show selected
+                    # radio buttons correctly. (It does if you remove the leading slash
+                    # from the value, but that causes them not to appear correctly in
+                    # Chrome/Firefox/Acrobat/etc).
+                    radio_fields_by_id[field_id]["radio_options"].append({
+                        "value": on_values[0],
+                        "rect": rect,
+                    })
+
+    # Some PDFs have form field definitions without corresponding annotations,
+    # so we can't tell where they are. Ignore these fields for now.
+    fields_with_location = []
+    for field_info in field_info_by_id.values():
+        if "page" in field_info:
+            fields_with_location.append(field_info)
+        else:
+            print(f"Unable to determine location for field id: {field_info.get('field_id')}, ignoring")
+
+    # Sort by page number, then Y position (flipped in PDF coordinate system), then X.
+    def sort_key(f):
+        if "radio_options" in f:
+            rect = f["radio_options"][0]["rect"] or [0, 0, 0, 0]
+        else:
+            rect = f.get("rect") or [0, 0, 0, 0]
+        adjusted_position = [-rect[1], rect[0]]
+        return [f.get("page"), adjusted_position]
+    
+    sorted_fields = fields_with_location + list(radio_fields_by_id.values())
+    sorted_fields.sort(key=sort_key)
+
+    return sorted_fields
+
+
+def write_field_info(pdf_path: str, json_output_path: str):
+    reader = PdfReader(pdf_path)
+    field_info = get_field_info(reader)
+    with open(json_output_path, "w") as f:
+        json.dump(field_info, f, indent=2)
+    print(f"Wrote {len(field_info)} fields to {json_output_path}")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 3:
+        print("Usage: extract_form_field_info.py [input pdf] [output json]")
+        sys.exit(1)
+    write_field_info(sys.argv[1], sys.argv[2])

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/fill_fillable_fields.py

@@ -0,0 +1,114 @@
+import json
+import sys
+
+from pypdf import PdfReader, PdfWriter
+
+from extract_form_field_info import get_field_info
+
+
+# Fills fillable form fields in a PDF. See forms.md.
+
+
+def fill_pdf_fields(input_pdf_path: str, fields_json_path: str, output_pdf_path: str):
+    with open(fields_json_path) as f:
+        fields = json.load(f)
+    # Group by page number.
+    fields_by_page = {}
+    for field in fields:
+        if "value" in field:
+            field_id = field["field_id"]
+            page = field["page"]
+            if page not in fields_by_page:
+                fields_by_page[page] = {}
+            fields_by_page[page][field_id] = field["value"]
+    
+    reader = PdfReader(input_pdf_path)
+
+    has_error = False
+    field_info = get_field_info(reader)
+    fields_by_ids = {f["field_id"]: f for f in field_info}
+    for field in fields:
+        existing_field = fields_by_ids.get(field["field_id"])
+        if not existing_field:
+            has_error = True
+            print(f"ERROR: `{field['field_id']}` is not a valid field ID")
+        elif field["page"] != existing_field["page"]:
+            has_error = True
+            print(f"ERROR: Incorrect page number for `{field['field_id']}` (got {field['page']}, expected {existing_field['page']})")
+        else:
+            if "value" in field:
+                err = validation_error_for_field_value(existing_field, field["value"])
+                if err:
+                    print(err)
+                    has_error = True
+    if has_error:
+        sys.exit(1)
+
+    writer = PdfWriter(clone_from=reader)
+    for page, field_values in fields_by_page.items():
+        writer.update_page_form_field_values(writer.pages[page - 1], field_values, auto_regenerate=False)
+
+    # This seems to be necessary for many PDF viewers to format the form values correctly.
+    # It may cause the viewer to show a "save changes" dialog even if the user doesn't make any changes.
+    writer.set_need_appearances_writer(True)
+    
+    with open(output_pdf_path, "wb") as f:
+        writer.write(f)
+
+
+def validation_error_for_field_value(field_info, field_value):
+    field_type = field_info["type"]
+    field_id = field_info["field_id"]
+    if field_type == "checkbox":
+        checked_val = field_info["checked_value"]
+        unchecked_val = field_info["unchecked_value"]
+        if field_value != checked_val and field_value != unchecked_val:
+            return f'ERROR: Invalid value "{field_value}" for checkbox field "{field_id}". The checked value is "{checked_val}" and the unchecked value is "{unchecked_val}"'
+    elif field_type == "radio_group":
+        option_values = [opt["value"] for opt in field_info["radio_options"]]
+        if field_value not in option_values:
+            return f'ERROR: Invalid value "{field_value}" for radio group field "{field_id}". Valid values are: {option_values}' 
+    elif field_type == "choice":
+        choice_values = [opt["value"] for opt in field_info["choice_options"]]
+        if field_value not in choice_values:
+            return f'ERROR: Invalid value "{field_value}" for choice field "{field_id}". Valid values are: {choice_values}'
+    return None
+
+
+# pypdf (at least version 5.7.0) has a bug when setting the value for a selection list field.
+# In _writer.py around line 966:
+#
+# if field.get(FA.FT, "/Tx") == "/Ch" and field_flags & FA.FfBits.Combo == 0:
+#     txt = "\n".join(annotation.get_inherited(FA.Opt, []))
+#
+# The problem is that for selection lists, `get_inherited` returns a list of two-element lists like
+# [["value1", "Text 1"], ["value2", "Text 2"], ...]
+# This causes `join` to throw a TypeError because it expects an iterable of strings.
+# The horrible workaround is to patch `get_inherited` to return a list of the value strings.
+# We call the original method and adjust the return value only if the argument to `get_inherited`
+# is `FA.Opt` and if the return value is a list of two-element lists.
+def monkeypatch_pydpf_method():
+    from pypdf.generic import DictionaryObject
+    from pypdf.constants import FieldDictionaryAttributes
+
+    original_get_inherited = DictionaryObject.get_inherited
+
+    def patched_get_inherited(self, key: str, default = None):
+        result = original_get_inherited(self, key, default)
+        if key == FieldDictionaryAttributes.Opt:
+            if isinstance(result, list) and all(isinstance(v, list) and len(v) == 2 for v in result):
+                result = [r[0] for r in result]
+        return result
+
+    DictionaryObject.get_inherited = patched_get_inherited
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 4:
+        print("Usage: fill_fillable_fields.py [input pdf] [field_values.json] [output pdf]")
+        sys.exit(1)
+    monkeypatch_pydpf_method()
+    input_pdf = sys.argv[1]
+    fields_json = sys.argv[2]
+    output_pdf = sys.argv[3]
+    fill_pdf_fields(input_pdf, fields_json, output_pdf)

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pdf/fill_pdf_form_with_annotations.py

@@ -0,0 +1,108 @@
+import json
+import sys
+
+from pypdf import PdfReader, PdfWriter
+from pypdf.annotations import FreeText
+
+
+# Fills a PDF by adding text annotations defined in `fields.json`. See forms.md.
+
+
+def transform_coordinates(bbox, image_width, image_height, pdf_width, pdf_height):
+    """Transform bounding box from image coordinates to PDF coordinates"""
+    # Image coordinates: origin at top-left, y increases downward
+    # PDF coordinates: origin at bottom-left, y increases upward
+    x_scale = pdf_width / image_width
+    y_scale = pdf_height / image_height
+    
+    left = bbox[0] * x_scale
+    right = bbox[2] * x_scale
+    
+    # Flip Y coordinates for PDF
+    top = pdf_height - (bbox[1] * y_scale)
+    bottom = pdf_height - (bbox[3] * y_scale)
+    
+    return left, bottom, right, top
+
+
+def fill_pdf_form(input_pdf_path, fields_json_path, output_pdf_path):
+    """Fill the PDF form with data from fields.json"""
+    
+    # `fields.json` format described in forms.md.
+    with open(fields_json_path, "r") as f:
+        fields_data = json.load(f)
+    
+    # Open the PDF
+    reader = PdfReader(input_pdf_path)
+    writer = PdfWriter()
+    
+    # Copy all pages to writer
+    writer.append(reader)
+    
+    # Get PDF dimensions for each page
+    pdf_dimensions = {}
+    for i, page in enumerate(reader.pages):
+        mediabox = page.mediabox
+        pdf_dimensions[i + 1] = [mediabox.width, mediabox.height]
+    
+    # Process each form field
+    annotations = []
+    for field in fields_data["form_fields"]:
+        page_num = field["page_number"]
+        
+        # Get page dimensions and transform coordinates.
+        page_info = next(p for p in fields_data["pages"] if p["page_number"] == page_num)
+        image_width = page_info["image_width"]
+        image_height = page_info["image_height"]
+        pdf_width, pdf_height = pdf_dimensions[page_num]
+        
+        transformed_entry_box = transform_coordinates(
+            field["entry_bounding_box"],
+            image_width, image_height,
+            pdf_width, pdf_height
+        )
+        
+        # Skip empty fields
+        if "entry_text" not in field or "text" not in field["entry_text"]:
+            continue
+        entry_text = field["entry_text"]
+        text = entry_text["text"]
+        if not text:
+            continue
+        
+        font_name = entry_text.get("font", "Arial")
+        font_size = str(entry_text.get("font_size", 14)) + "pt"
+        font_color = entry_text.get("font_color", "000000")
+
+        # Font size/color seems to not work reliably across viewers:
+        # https://github.com/py-pdf/pypdf/issues/2084
+        annotation = FreeText(
+            text=text,
+            rect=transformed_entry_box,
+            font=font_name,
+            font_size=font_size,
+            font_color=font_color,
+            border_color=None,
+            background_color=None,
+        )
+        annotations.append(annotation)
+        # page_number is 0-based for pypdf
+        writer.add_annotation(page_number=page_num - 1, annotation=annotation)
+        
+    # Save the filled PDF
+    with open(output_pdf_path, "wb") as output:
+        writer.write(output)
+    
+    print(f"Successfully filled PDF form and saved to {output_pdf_path}")
+    print(f"Added {len(annotations)} text annotations")
+
+
+if __name__ == "__main__":
+    if len(sys.argv) != 4:
+        print("Usage: fill_pdf_form_with_annotations.py [input pdf] [fields.json] [output pdf]")
+        sys.exit(1)
+    input_pdf = sys.argv[1]
+    fields_json = sys.argv[2]
+    output_pdf = sys.argv[3]
+    
+    fill_pdf_form(input_pdf, fields_json, output_pdf)

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pptx/rearrange.py

@@ -0,0 +1,231 @@
+#!/usr/bin/env python3
+"""
+Rearrange PowerPoint slides based on a sequence of indices.
+
+Usage:
+    python rearrange.py template.pptx output.pptx 0,34,34,50,52
+
+This will create output.pptx using slides from template.pptx in the specified order.
+Slides can be repeated (e.g., 34 appears twice).
+"""
+
+import argparse
+import shutil
+import sys
+from copy import deepcopy
+from pathlib import Path
+
+import six
+from pptx import Presentation
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Rearrange PowerPoint slides based on a sequence of indices.",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  python rearrange.py template.pptx output.pptx 0,34,34,50,52
+    Creates output.pptx using slides 0, 34 (twice), 50, and 52 from template.pptx
+
+  python rearrange.py template.pptx output.pptx 5,3,1,2,4
+    Creates output.pptx with slides reordered as specified
+
+Note: Slide indices are 0-based (first slide is 0, second is 1, etc.)
+        """,
+    )
+
+    parser.add_argument("template", help="Path to template PPTX file")
+    parser.add_argument("output", help="Path for output PPTX file")
+    parser.add_argument(
+        "sequence", help="Comma-separated sequence of slide indices (0-based)"
+    )
+
+    args = parser.parse_args()
+
+    # Parse the slide sequence
+    try:
+        slide_sequence = [int(x.strip()) for x in args.sequence.split(",")]
+    except ValueError:
+        print(
+            "Error: Invalid sequence format. Use comma-separated integers (e.g., 0,34,34,50,52)"
+        )
+        sys.exit(1)
+
+    # Check template exists
+    template_path = Path(args.template)
+    if not template_path.exists():
+        print(f"Error: Template file not found: {args.template}")
+        sys.exit(1)
+
+    # Create output directory if needed
+    output_path = Path(args.output)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    try:
+        rearrange_presentation(template_path, output_path, slide_sequence)
+    except ValueError as e:
+        print(f"Error: {e}")
+        sys.exit(1)
+    except Exception as e:
+        print(f"Error processing presentation: {e}")
+        sys.exit(1)
+
+
+def duplicate_slide(pres, index):
+    """Duplicate a slide in the presentation."""
+    source = pres.slides[index]
+
+    # Use source's layout to preserve formatting
+    new_slide = pres.slides.add_slide(source.slide_layout)
+
+    # Collect all image and media relationships from the source slide
+    image_rels = {}
+    for rel_id, rel in six.iteritems(source.part.rels):
+        if "image" in rel.reltype or "media" in rel.reltype:
+            image_rels[rel_id] = rel
+
+    # CRITICAL: Clear placeholder shapes to avoid duplicates
+    for shape in new_slide.shapes:
+        sp = shape.element
+        sp.getparent().remove(sp)
+
+    # Copy all shapes from source
+    for shape in source.shapes:
+        el = shape.element
+        new_el = deepcopy(el)
+        new_slide.shapes._spTree.insert_element_before(new_el, "p:extLst")
+
+        # Handle picture shapes - need to update the blip reference
+        # Look for all blip elements (they can be in pic or other contexts)
+        # Using the element's own xpath method without namespaces argument
+        blips = new_el.xpath(".//a:blip[@r:embed]")
+        for blip in blips:
+            old_rId = blip.get(
+                "{http://schemas.openxmlformats.org/officeDocument/2006/relationships}embed"
+            )
+            if old_rId in image_rels:
+                # Create a new relationship in the destination slide for this image
+                old_rel = image_rels[old_rId]
+                # get_or_add returns the rId directly, or adds and returns new rId
+                new_rId = new_slide.part.rels.get_or_add(
+                    old_rel.reltype, old_rel._target
+                )
+                # Update the blip's embed reference to use the new relationship ID
+                blip.set(
+                    "{http://schemas.openxmlformats.org/officeDocument/2006/relationships}embed",
+                    new_rId,
+                )
+
+    # Copy any additional image/media relationships that might be referenced elsewhere
+    for rel_id, rel in image_rels.items():
+        try:
+            new_slide.part.rels.get_or_add(rel.reltype, rel._target)
+        except Exception:
+            pass  # Relationship might already exist
+
+    return new_slide
+
+
+def delete_slide(pres, index):
+    """Delete a slide from the presentation."""
+    rId = pres.slides._sldIdLst[index].rId
+    pres.part.drop_rel(rId)
+    del pres.slides._sldIdLst[index]
+
+
+def reorder_slides(pres, slide_index, target_index):
+    """Move a slide from one position to another."""
+    slides = pres.slides._sldIdLst
+
+    # Remove slide element from current position
+    slide_element = slides[slide_index]
+    slides.remove(slide_element)
+
+    # Insert at target position
+    slides.insert(target_index, slide_element)
+
+
+def rearrange_presentation(template_path, output_path, slide_sequence):
+    """
+    Create a new presentation with slides from template in specified order.
+
+    Args:
+        template_path: Path to template PPTX file
+        output_path: Path for output PPTX file
+        slide_sequence: List of slide indices (0-based) to include
+    """
+    # Copy template to preserve dimensions and theme
+    if template_path != output_path:
+        shutil.copy2(template_path, output_path)
+        prs = Presentation(output_path)
+    else:
+        prs = Presentation(template_path)
+
+    total_slides = len(prs.slides)
+
+    # Validate indices
+    for idx in slide_sequence:
+        if idx < 0 or idx >= total_slides:
+            raise ValueError(f"Slide index {idx} out of range (0-{total_slides - 1})")
+
+    # Track original slides and their duplicates
+    slide_map = []  # List of actual slide indices for final presentation
+    duplicated = {}  # Track duplicates: original_idx -> [duplicate_indices]
+
+    # Step 1: DUPLICATE repeated slides
+    print(f"Processing {len(slide_sequence)} slides from template...")
+    for i, template_idx in enumerate(slide_sequence):
+        if template_idx in duplicated and duplicated[template_idx]:
+            # Already duplicated this slide, use the duplicate
+            slide_map.append(duplicated[template_idx].pop(0))
+            print(f"  [{i}] Using duplicate of slide {template_idx}")
+        elif slide_sequence.count(template_idx) > 1 and template_idx not in duplicated:
+            # First occurrence of a repeated slide - create duplicates
+            slide_map.append(template_idx)
+            duplicates = []
+            count = slide_sequence.count(template_idx) - 1
+            print(
+                f"  [{i}] Using original slide {template_idx}, creating {count} duplicate(s)"
+            )
+            for _ in range(count):
+                duplicate_slide(prs, template_idx)
+                duplicates.append(len(prs.slides) - 1)
+            duplicated[template_idx] = duplicates
+        else:
+            # Unique slide or first occurrence already handled, use original
+            slide_map.append(template_idx)
+            print(f"  [{i}] Using original slide {template_idx}")
+
+    # Step 2: DELETE unwanted slides (work backwards)
+    slides_to_keep = set(slide_map)
+    print(f"\nDeleting {len(prs.slides) - len(slides_to_keep)} unused slides...")
+    for i in range(len(prs.slides) - 1, -1, -1):
+        if i not in slides_to_keep:
+            delete_slide(prs, i)
+            # Update slide_map indices after deletion
+            slide_map = [idx - 1 if idx > i else idx for idx in slide_map]
+
+    # Step 3: REORDER to final sequence
+    print(f"Reordering {len(slide_map)} slides to final sequence...")
+    for target_pos in range(len(slide_map)):
+        # Find which slide should be at target_pos
+        current_pos = slide_map[target_pos]
+        if current_pos != target_pos:
+            reorder_slides(prs, current_pos, target_pos)
+            # Update slide_map: the move shifts other slides
+            for i in range(len(slide_map)):
+                if slide_map[i] > current_pos and slide_map[i] <= target_pos:
+                    slide_map[i] -= 1
+                elif slide_map[i] < current_pos and slide_map[i] >= target_pos:
+                    slide_map[i] += 1
+            slide_map[target_pos] = target_pos
+
+    # Save the presentation
+    prs.save(output_path)
+    print(f"\nSaved rearranged presentation to: {output_path}")
+    print(f"Final presentation has {len(prs.slides)} slides")
+
+
+if __name__ == "__main__":
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pptx/replace.py

@@ -0,0 +1,385 @@
+#!/usr/bin/env python3
+"""Apply text replacements to PowerPoint presentation.
+
+Usage:
+    python replace.py <input.pptx> <replacements.json> <output.pptx>
+
+The replacements JSON should have the structure output by inventory.py.
+ALL text shapes identified by inventory.py will have their text cleared
+unless "paragraphs" is specified in the replacements for that shape.
+"""
+
+import json
+import sys
+from pathlib import Path
+from typing import Any, Dict, List
+
+from inventory import InventoryData, extract_text_inventory
+from pptx import Presentation
+from pptx.dml.color import RGBColor
+from pptx.enum.dml import MSO_THEME_COLOR
+from pptx.enum.text import PP_ALIGN
+from pptx.oxml.xmlchemy import OxmlElement
+from pptx.util import Pt
+
+
+def clear_paragraph_bullets(paragraph):
+    """Clear bullet formatting from a paragraph."""
+    pPr = paragraph._element.get_or_add_pPr()
+
+    # Remove existing bullet elements
+    for child in list(pPr):
+        if (
+            child.tag.endswith("buChar")
+            or child.tag.endswith("buNone")
+            or child.tag.endswith("buAutoNum")
+            or child.tag.endswith("buFont")
+        ):
+            pPr.remove(child)
+
+    return pPr
+
+
+def apply_paragraph_properties(paragraph, para_data: Dict[str, Any]):
+    """Apply formatting properties to a paragraph."""
+    # Get the text but don't set it on paragraph directly yet
+    text = para_data.get("text", "")
+
+    # Get or create paragraph properties
+    pPr = clear_paragraph_bullets(paragraph)
+
+    # Handle bullet formatting
+    if para_data.get("bullet", False):
+        level = para_data.get("level", 0)
+        paragraph.level = level
+
+        # Calculate font-proportional indentation
+        font_size = para_data.get("font_size", 18.0)
+        level_indent_emu = int((font_size * (1.6 + level * 1.6)) * 12700)
+        hanging_indent_emu = int(-font_size * 0.8 * 12700)
+
+        # Set indentation
+        pPr.attrib["marL"] = str(level_indent_emu)
+        pPr.attrib["indent"] = str(hanging_indent_emu)
+
+        # Add bullet character
+        buChar = OxmlElement("a:buChar")
+        buChar.set("char", "•")
+        pPr.append(buChar)
+
+        # Default to left alignment for bullets if not specified
+        if "alignment" not in para_data:
+            paragraph.alignment = PP_ALIGN.LEFT
+    else:
+        # Remove indentation for non-bullet text
+        pPr.attrib["marL"] = "0"
+        pPr.attrib["indent"] = "0"
+
+        # Add buNone element
+        buNone = OxmlElement("a:buNone")
+        pPr.insert(0, buNone)
+
+    # Apply alignment
+    if "alignment" in para_data:
+        alignment_map = {
+            "LEFT": PP_ALIGN.LEFT,
+            "CENTER": PP_ALIGN.CENTER,
+            "RIGHT": PP_ALIGN.RIGHT,
+            "JUSTIFY": PP_ALIGN.JUSTIFY,
+        }
+        if para_data["alignment"] in alignment_map:
+            paragraph.alignment = alignment_map[para_data["alignment"]]
+
+    # Apply spacing
+    if "space_before" in para_data:
+        paragraph.space_before = Pt(para_data["space_before"])
+    if "space_after" in para_data:
+        paragraph.space_after = Pt(para_data["space_after"])
+    if "line_spacing" in para_data:
+        paragraph.line_spacing = Pt(para_data["line_spacing"])
+
+    # Apply run-level formatting
+    if not paragraph.runs:
+        run = paragraph.add_run()
+        run.text = text
+    else:
+        run = paragraph.runs[0]
+        run.text = text
+
+    # Apply font properties
+    apply_font_properties(run, para_data)
+
+
+def apply_font_properties(run, para_data: Dict[str, Any]):
+    """Apply font properties to a text run."""
+    if "bold" in para_data:
+        run.font.bold = para_data["bold"]
+    if "italic" in para_data:
+        run.font.italic = para_data["italic"]
+    if "underline" in para_data:
+        run.font.underline = para_data["underline"]
+    if "font_size" in para_data:
+        run.font.size = Pt(para_data["font_size"])
+    if "font_name" in para_data:
+        run.font.name = para_data["font_name"]
+
+    # Apply color - prefer RGB, fall back to theme_color
+    if "color" in para_data:
+        color_hex = para_data["color"].lstrip("#")
+        if len(color_hex) == 6:
+            r = int(color_hex[0:2], 16)
+            g = int(color_hex[2:4], 16)
+            b = int(color_hex[4:6], 16)
+            run.font.color.rgb = RGBColor(r, g, b)
+    elif "theme_color" in para_data:
+        # Get theme color by name (e.g., "DARK_1", "ACCENT_1")
+        theme_name = para_data["theme_color"]
+        try:
+            run.font.color.theme_color = getattr(MSO_THEME_COLOR, theme_name)
+        except AttributeError:
+            print(f"  WARNING: Unknown theme color name '{theme_name}'")
+
+
+def detect_frame_overflow(inventory: InventoryData) -> Dict[str, Dict[str, float]]:
+    """Detect text overflow in shapes (text exceeding shape bounds).
+
+    Returns dict of slide_key -> shape_key -> overflow_inches.
+    Only includes shapes that have text overflow.
+    """
+    overflow_map = {}
+
+    for slide_key, shapes_dict in inventory.items():
+        for shape_key, shape_data in shapes_dict.items():
+            # Check for frame overflow (text exceeding shape bounds)
+            if shape_data.frame_overflow_bottom is not None:
+                if slide_key not in overflow_map:
+                    overflow_map[slide_key] = {}
+                overflow_map[slide_key][shape_key] = shape_data.frame_overflow_bottom
+
+    return overflow_map
+
+
+def validate_replacements(inventory: InventoryData, replacements: Dict) -> List[str]:
+    """Validate that all shapes in replacements exist in inventory.
+
+    Returns list of error messages.
+    """
+    errors = []
+
+    for slide_key, shapes_data in replacements.items():
+        if not slide_key.startswith("slide-"):
+            continue
+
+        # Check if slide exists
+        if slide_key not in inventory:
+            errors.append(f"Slide '{slide_key}' not found in inventory")
+            continue
+
+        # Check each shape
+        for shape_key in shapes_data.keys():
+            if shape_key not in inventory[slide_key]:
+                # Find shapes without replacements defined and show their content
+                unused_with_content = []
+                for k in inventory[slide_key].keys():
+                    if k not in shapes_data:
+                        shape_data = inventory[slide_key][k]
+                        # Get text from paragraphs as preview
+                        paragraphs = shape_data.paragraphs
+                        if paragraphs and paragraphs[0].text:
+                            first_text = paragraphs[0].text[:50]
+                            if len(paragraphs[0].text) > 50:
+                                first_text += "..."
+                            unused_with_content.append(f"{k} ('{first_text}')")
+                        else:
+                            unused_with_content.append(k)
+
+                errors.append(
+                    f"Shape '{shape_key}' not found on '{slide_key}'. "
+                    f"Shapes without replacements: {', '.join(sorted(unused_with_content)) if unused_with_content else 'none'}"
+                )
+
+    return errors
+
+
+def check_duplicate_keys(pairs):
+    """Check for duplicate keys when loading JSON."""
+    result = {}
+    for key, value in pairs:
+        if key in result:
+            raise ValueError(f"Duplicate key found in JSON: '{key}'")
+        result[key] = value
+    return result
+
+
+def apply_replacements(pptx_file: str, json_file: str, output_file: str):
+    """Apply text replacements from JSON to PowerPoint presentation."""
+
+    # Load presentation
+    prs = Presentation(pptx_file)
+
+    # Get inventory of all text shapes (returns ShapeData objects)
+    # Pass prs to use same Presentation instance
+    inventory = extract_text_inventory(Path(pptx_file), prs)
+
+    # Detect text overflow in original presentation
+    original_overflow = detect_frame_overflow(inventory)
+
+    # Load replacement data with duplicate key detection
+    with open(json_file, "r") as f:
+        replacements = json.load(f, object_pairs_hook=check_duplicate_keys)
+
+    # Validate replacements
+    errors = validate_replacements(inventory, replacements)
+    if errors:
+        print("ERROR: Invalid shapes in replacement JSON:")
+        for error in errors:
+            print(f"  - {error}")
+        print("\nPlease check the inventory and update your replacement JSON.")
+        print(
+            "You can regenerate the inventory with: python inventory.py <input.pptx> <output.json>"
+        )
+        raise ValueError(f"Found {len(errors)} validation error(s)")
+
+    # Track statistics
+    shapes_processed = 0
+    shapes_cleared = 0
+    shapes_replaced = 0
+
+    # Process each slide from inventory
+    for slide_key, shapes_dict in inventory.items():
+        if not slide_key.startswith("slide-"):
+            continue
+
+        slide_index = int(slide_key.split("-")[1])
+
+        if slide_index >= len(prs.slides):
+            print(f"Warning: Slide {slide_index} not found")
+            continue
+
+        # Process each shape from inventory
+        for shape_key, shape_data in shapes_dict.items():
+            shapes_processed += 1
+
+            # Get the shape directly from ShapeData
+            shape = shape_data.shape
+            if not shape:
+                print(f"Warning: {shape_key} has no shape reference")
+                continue
+
+            # ShapeData already validates text_frame in __init__
+            text_frame = shape.text_frame  # type: ignore
+
+            text_frame.clear()  # type: ignore
+            shapes_cleared += 1
+
+            # Check for replacement paragraphs
+            replacement_shape_data = replacements.get(slide_key, {}).get(shape_key, {})
+            if "paragraphs" not in replacement_shape_data:
+                continue
+
+            shapes_replaced += 1
+
+            # Add replacement paragraphs
+            for i, para_data in enumerate(replacement_shape_data["paragraphs"]):
+                if i == 0:
+                    p = text_frame.paragraphs[0]  # type: ignore
+                else:
+                    p = text_frame.add_paragraph()  # type: ignore
+
+                apply_paragraph_properties(p, para_data)
+
+    # Check for issues after replacements
+    # Save to a temporary file and reload to avoid modifying the presentation during inventory
+    # (extract_text_inventory accesses font.color which adds empty <a:solidFill/> elements)
+    import tempfile
+
+    with tempfile.NamedTemporaryFile(suffix=".pptx", delete=False) as tmp:
+        tmp_path = Path(tmp.name)
+        prs.save(str(tmp_path))
+
+    try:
+        updated_inventory = extract_text_inventory(tmp_path)
+        updated_overflow = detect_frame_overflow(updated_inventory)
+    finally:
+        tmp_path.unlink()  # Clean up temp file
+
+    # Check if any text overflow got worse
+    overflow_errors = []
+    for slide_key, shape_overflows in updated_overflow.items():
+        for shape_key, new_overflow in shape_overflows.items():
+            # Get original overflow (0 if there was no overflow before)
+            original = original_overflow.get(slide_key, {}).get(shape_key, 0.0)
+
+            # Error if overflow increased
+            if new_overflow > original + 0.01:  # Small tolerance for rounding
+                increase = new_overflow - original
+                overflow_errors.append(
+                    f'{slide_key}/{shape_key}: overflow worsened by {increase:.2f}" '
+                    f'(was {original:.2f}", now {new_overflow:.2f}")'
+                )
+
+    # Collect warnings from updated shapes
+    warnings = []
+    for slide_key, shapes_dict in updated_inventory.items():
+        for shape_key, shape_data in shapes_dict.items():
+            if shape_data.warnings:
+                for warning in shape_data.warnings:
+                    warnings.append(f"{slide_key}/{shape_key}: {warning}")
+
+    # Fail if there are any issues
+    if overflow_errors or warnings:
+        print("\nERROR: Issues detected in replacement output:")
+        if overflow_errors:
+            print("\nText overflow worsened:")
+            for error in overflow_errors:
+                print(f"  - {error}")
+        if warnings:
+            print("\nFormatting warnings:")
+            for warning in warnings:
+                print(f"  - {warning}")
+        print("\nPlease fix these issues before saving.")
+        raise ValueError(
+            f"Found {len(overflow_errors)} overflow error(s) and {len(warnings)} warning(s)"
+        )
+
+    # Save the presentation
+    prs.save(output_file)
+
+    # Report results
+    print(f"Saved updated presentation to: {output_file}")
+    print(f"Processed {len(prs.slides)} slides")
+    print(f"  - Shapes processed: {shapes_processed}")
+    print(f"  - Shapes cleared: {shapes_cleared}")
+    print(f"  - Shapes replaced: {shapes_replaced}")
+
+
+def main():
+    """Main entry point for command-line usage."""
+    if len(sys.argv) != 4:
+        print(__doc__)
+        sys.exit(1)
+
+    input_pptx = Path(sys.argv[1])
+    replacements_json = Path(sys.argv[2])
+    output_pptx = Path(sys.argv[3])
+
+    if not input_pptx.exists():
+        print(f"Error: Input file '{input_pptx}' not found")
+        sys.exit(1)
+
+    if not replacements_json.exists():
+        print(f"Error: Replacements JSON file '{replacements_json}' not found")
+        sys.exit(1)
+
+    try:
+        apply_replacements(str(input_pptx), str(replacements_json), str(output_pptx))
+    except Exception as e:
+        print(f"Error applying replacements: {e}")
+        import traceback
+
+        traceback.print_exc()
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/pptx/thumbnail.py

@@ -0,0 +1,450 @@
+#!/usr/bin/env python3
+"""
+Create thumbnail grids from PowerPoint presentation slides.
+
+Creates a grid layout of slide thumbnails with configurable columns (max 6).
+Each grid contains up to cols×(cols+1) images. For presentations with more
+slides, multiple numbered grid files are created automatically.
+
+The program outputs the names of all files created.
+
+Output:
+- Single grid: {prefix}.jpg (if slides fit in one grid)
+- Multiple grids: {prefix}-1.jpg, {prefix}-2.jpg, etc.
+
+Grid limits by column count:
+- 3 cols: max 12 slides per grid (3×4)
+- 4 cols: max 20 slides per grid (4×5)
+- 5 cols: max 30 slides per grid (5×6) [default]
+- 6 cols: max 42 slides per grid (6×7)
+
+Usage:
+    python thumbnail.py input.pptx [output_prefix] [--cols N] [--outline-placeholders]
+
+Examples:
+    python thumbnail.py presentation.pptx
+    # Creates: thumbnails.jpg (using default prefix)
+    # Outputs:
+    #   Created 1 grid(s):
+    #     - thumbnails.jpg
+
+    python thumbnail.py large-deck.pptx grid --cols 4
+    # Creates: grid-1.jpg, grid-2.jpg, grid-3.jpg
+    # Outputs:
+    #   Created 3 grid(s):
+    #     - grid-1.jpg
+    #     - grid-2.jpg
+    #     - grid-3.jpg
+
+    python thumbnail.py template.pptx analysis --outline-placeholders
+    # Creates thumbnail grids with red outlines around text placeholders
+"""
+
+import argparse
+import subprocess
+import sys
+import tempfile
+from pathlib import Path
+
+from inventory import extract_text_inventory
+from PIL import Image, ImageDraw, ImageFont
+from pptx import Presentation
+
+# Constants
+THUMBNAIL_WIDTH = 300  # Fixed thumbnail width in pixels
+CONVERSION_DPI = 100  # DPI for PDF to image conversion
+MAX_COLS = 6  # Maximum number of columns
+DEFAULT_COLS = 5  # Default number of columns
+JPEG_QUALITY = 95  # JPEG compression quality
+
+# Grid layout constants
+GRID_PADDING = 20  # Padding between thumbnails
+BORDER_WIDTH = 2  # Border width around thumbnails
+FONT_SIZE_RATIO = 0.12  # Font size as fraction of thumbnail width
+LABEL_PADDING_RATIO = 0.4  # Label padding as fraction of font size
+
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Create thumbnail grids from PowerPoint slides."
+    )
+    parser.add_argument("input", help="Input PowerPoint file (.pptx)")
+    parser.add_argument(
+        "output_prefix",
+        nargs="?",
+        default="thumbnails",
+        help="Output prefix for image files (default: thumbnails, will create prefix.jpg or prefix-N.jpg)",
+    )
+    parser.add_argument(
+        "--cols",
+        type=int,
+        default=DEFAULT_COLS,
+        help=f"Number of columns (default: {DEFAULT_COLS}, max: {MAX_COLS})",
+    )
+    parser.add_argument(
+        "--outline-placeholders",
+        action="store_true",
+        help="Outline text placeholders with a colored border",
+    )
+
+    args = parser.parse_args()
+
+    # Validate columns
+    cols = min(args.cols, MAX_COLS)
+    if args.cols > MAX_COLS:
+        print(f"Warning: Columns limited to {MAX_COLS} (requested {args.cols})")
+
+    # Validate input
+    input_path = Path(args.input)
+    if not input_path.exists() or input_path.suffix.lower() != ".pptx":
+        print(f"Error: Invalid PowerPoint file: {args.input}")
+        sys.exit(1)
+
+    # Construct output path (always JPG)
+    output_path = Path(f"{args.output_prefix}.jpg")
+
+    print(f"Processing: {args.input}")
+
+    try:
+        with tempfile.TemporaryDirectory() as temp_dir:
+            # Get placeholder regions if outlining is enabled
+            placeholder_regions = None
+            slide_dimensions = None
+            if args.outline_placeholders:
+                print("Extracting placeholder regions...")
+                placeholder_regions, slide_dimensions = get_placeholder_regions(
+                    input_path
+                )
+                if placeholder_regions:
+                    print(f"Found placeholders on {len(placeholder_regions)} slides")
+
+            # Convert slides to images
+            slide_images = convert_to_images(input_path, Path(temp_dir), CONVERSION_DPI)
+            if not slide_images:
+                print("Error: No slides found")
+                sys.exit(1)
+
+            print(f"Found {len(slide_images)} slides")
+
+            # Create grids (max cols×(cols+1) images per grid)
+            grid_files = create_grids(
+                slide_images,
+                cols,
+                THUMBNAIL_WIDTH,
+                output_path,
+                placeholder_regions,
+                slide_dimensions,
+            )
+
+            # Print saved files
+            print(f"Created {len(grid_files)} grid(s):")
+            for grid_file in grid_files:
+                print(f"  - {grid_file}")
+
+    except Exception as e:
+        print(f"Error: {e}")
+        sys.exit(1)
+
+
+def create_hidden_slide_placeholder(size):
+    """Create placeholder image for hidden slides."""
+    img = Image.new("RGB", size, color="#F0F0F0")
+    draw = ImageDraw.Draw(img)
+    line_width = max(5, min(size) // 100)
+    draw.line([(0, 0), size], fill="#CCCCCC", width=line_width)
+    draw.line([(size[0], 0), (0, size[1])], fill="#CCCCCC", width=line_width)
+    return img
+
+
+def get_placeholder_regions(pptx_path):
+    """Extract ALL text regions from the presentation.
+
+    Returns a tuple of (placeholder_regions, slide_dimensions).
+    text_regions is a dict mapping slide indices to lists of text regions.
+    Each region is a dict with 'left', 'top', 'width', 'height' in inches.
+    slide_dimensions is a tuple of (width_inches, height_inches).
+    """
+    prs = Presentation(str(pptx_path))
+    inventory = extract_text_inventory(pptx_path, prs)
+    placeholder_regions = {}
+
+    # Get actual slide dimensions in inches (EMU to inches conversion)
+    slide_width_inches = (prs.slide_width or 9144000) / 914400.0
+    slide_height_inches = (prs.slide_height or 5143500) / 914400.0
+
+    for slide_key, shapes in inventory.items():
+        # Extract slide index from "slide-N" format
+        slide_idx = int(slide_key.split("-")[1])
+        regions = []
+
+        for shape_key, shape_data in shapes.items():
+            # The inventory only contains shapes with text, so all shapes should be highlighted
+            regions.append(
+                {
+                    "left": shape_data.left,
+                    "top": shape_data.top,
+                    "width": shape_data.width,
+                    "height": shape_data.height,
+                }
+            )
+
+        if regions:
+            placeholder_regions[slide_idx] = regions
+
+    return placeholder_regions, (slide_width_inches, slide_height_inches)
+
+
+def convert_to_images(pptx_path, temp_dir, dpi):
+    """Convert PowerPoint to images via PDF, handling hidden slides."""
+    # Detect hidden slides
+    print("Analyzing presentation...")
+    prs = Presentation(str(pptx_path))
+    total_slides = len(prs.slides)
+
+    # Find hidden slides (1-based indexing for display)
+    hidden_slides = {
+        idx + 1
+        for idx, slide in enumerate(prs.slides)
+        if slide.element.get("show") == "0"
+    }
+
+    print(f"Total slides: {total_slides}")
+    if hidden_slides:
+        print(f"Hidden slides: {sorted(hidden_slides)}")
+
+    pdf_path = temp_dir / f"{pptx_path.stem}.pdf"
+
+    # Convert to PDF
+    print("Converting to PDF...")
+    result = subprocess.run(
+        [
+            "soffice",
+            "--headless",
+            "--convert-to",
+            "pdf",
+            "--outdir",
+            str(temp_dir),
+            str(pptx_path),
+        ],
+        capture_output=True,
+        text=True,
+    )
+    if result.returncode != 0 or not pdf_path.exists():
+        raise RuntimeError("PDF conversion failed")
+
+    # Convert PDF to images
+    print(f"Converting to images at {dpi} DPI...")
+    result = subprocess.run(
+        ["pdftoppm", "-jpeg", "-r", str(dpi), str(pdf_path), str(temp_dir / "slide")],
+        capture_output=True,
+        text=True,
+    )
+    if result.returncode != 0:
+        raise RuntimeError("Image conversion failed")
+
+    visible_images = sorted(temp_dir.glob("slide-*.jpg"))
+
+    # Create full list with placeholders for hidden slides
+    all_images = []
+    visible_idx = 0
+
+    # Get placeholder dimensions from first visible slide
+    if visible_images:
+        with Image.open(visible_images[0]) as img:
+            placeholder_size = img.size
+    else:
+        placeholder_size = (1920, 1080)
+
+    for slide_num in range(1, total_slides + 1):
+        if slide_num in hidden_slides:
+            # Create placeholder image for hidden slide
+            placeholder_path = temp_dir / f"hidden-{slide_num:03d}.jpg"
+            placeholder_img = create_hidden_slide_placeholder(placeholder_size)
+            placeholder_img.save(placeholder_path, "JPEG")
+            all_images.append(placeholder_path)
+        else:
+            # Use the actual visible slide image
+            if visible_idx < len(visible_images):
+                all_images.append(visible_images[visible_idx])
+                visible_idx += 1
+
+    return all_images
+
+
+def create_grids(
+    image_paths,
+    cols,
+    width,
+    output_path,
+    placeholder_regions=None,
+    slide_dimensions=None,
+):
+    """Create multiple thumbnail grids from slide images, max cols×(cols+1) images per grid."""
+    # Maximum images per grid is cols × (cols + 1) for better proportions
+    max_images_per_grid = cols * (cols + 1)
+    grid_files = []
+
+    print(
+        f"Creating grids with {cols} columns (max {max_images_per_grid} images per grid)"
+    )
+
+    # Split images into chunks
+    for chunk_idx, start_idx in enumerate(
+        range(0, len(image_paths), max_images_per_grid)
+    ):
+        end_idx = min(start_idx + max_images_per_grid, len(image_paths))
+        chunk_images = image_paths[start_idx:end_idx]
+
+        # Create grid for this chunk
+        grid = create_grid(
+            chunk_images, cols, width, start_idx, placeholder_regions, slide_dimensions
+        )
+
+        # Generate output filename
+        if len(image_paths) <= max_images_per_grid:
+            # Single grid - use base filename without suffix
+            grid_filename = output_path
+        else:
+            # Multiple grids - insert index before extension with dash
+            stem = output_path.stem
+            suffix = output_path.suffix
+            grid_filename = output_path.parent / f"{stem}-{chunk_idx + 1}{suffix}"
+
+        # Save grid
+        grid_filename.parent.mkdir(parents=True, exist_ok=True)
+        grid.save(str(grid_filename), quality=JPEG_QUALITY)
+        grid_files.append(str(grid_filename))
+
+    return grid_files
+
+
+def create_grid(
+    image_paths,
+    cols,
+    width,
+    start_slide_num=0,
+    placeholder_regions=None,
+    slide_dimensions=None,
+):
+    """Create thumbnail grid from slide images with optional placeholder outlining."""
+    font_size = int(width * FONT_SIZE_RATIO)
+    label_padding = int(font_size * LABEL_PADDING_RATIO)
+
+    # Get dimensions
+    with Image.open(image_paths[0]) as img:
+        aspect = img.height / img.width
+    height = int(width * aspect)
+
+    # Calculate grid size
+    rows = (len(image_paths) + cols - 1) // cols
+    grid_w = cols * width + (cols + 1) * GRID_PADDING
+    grid_h = rows * (height + font_size + label_padding * 2) + (rows + 1) * GRID_PADDING
+
+    # Create grid
+    grid = Image.new("RGB", (grid_w, grid_h), "white")
+    draw = ImageDraw.Draw(grid)
+
+    # Load font with size based on thumbnail width
+    try:
+        # Use Pillow's default font with size
+        font = ImageFont.load_default(size=font_size)
+    except Exception:
+        # Fall back to basic default font if size parameter not supported
+        font = ImageFont.load_default()
+
+    # Place thumbnails
+    for i, img_path in enumerate(image_paths):
+        row, col = i // cols, i % cols
+        x = col * width + (col + 1) * GRID_PADDING
+        y_base = (
+            row * (height + font_size + label_padding * 2) + (row + 1) * GRID_PADDING
+        )
+
+        # Add label with actual slide number
+        label = f"{start_slide_num + i}"
+        bbox = draw.textbbox((0, 0), label, font=font)
+        text_w = bbox[2] - bbox[0]
+        draw.text(
+            (x + (width - text_w) // 2, y_base + label_padding),
+            label,
+            fill="black",
+            font=font,
+        )
+
+        # Add thumbnail below label with proportional spacing
+        y_thumbnail = y_base + label_padding + font_size + label_padding
+
+        with Image.open(img_path) as img:
+            # Get original dimensions before thumbnail
+            orig_w, orig_h = img.size
+
+            # Apply placeholder outlines if enabled
+            if placeholder_regions and (start_slide_num + i) in placeholder_regions:
+                # Convert to RGBA for transparency support
+                if img.mode != "RGBA":
+                    img = img.convert("RGBA")
+
+                # Get the regions for this slide
+                regions = placeholder_regions[start_slide_num + i]
+
+                # Calculate scale factors using actual slide dimensions
+                if slide_dimensions:
+                    slide_width_inches, slide_height_inches = slide_dimensions
+                else:
+                    # Fallback: estimate from image size at CONVERSION_DPI
+                    slide_width_inches = orig_w / CONVERSION_DPI
+                    slide_height_inches = orig_h / CONVERSION_DPI
+
+                x_scale = orig_w / slide_width_inches
+                y_scale = orig_h / slide_height_inches
+
+                # Create a highlight overlay
+                overlay = Image.new("RGBA", img.size, (255, 255, 255, 0))
+                overlay_draw = ImageDraw.Draw(overlay)
+
+                # Highlight each placeholder region
+                for region in regions:
+                    # Convert from inches to pixels in the original image
+                    px_left = int(region["left"] * x_scale)
+                    px_top = int(region["top"] * y_scale)
+                    px_width = int(region["width"] * x_scale)
+                    px_height = int(region["height"] * y_scale)
+
+                    # Draw highlight outline with red color and thick stroke
+                    # Using a bright red outline instead of fill
+                    stroke_width = max(
+                        5, min(orig_w, orig_h) // 150
+                    )  # Thicker proportional stroke width
+                    overlay_draw.rectangle(
+                        [(px_left, px_top), (px_left + px_width, px_top + px_height)],
+                        outline=(255, 0, 0, 255),  # Bright red, fully opaque
+                        width=stroke_width,
+                    )
+
+                # Composite the overlay onto the image using alpha blending
+                img = Image.alpha_composite(img, overlay)
+                # Convert back to RGB for JPEG saving
+                img = img.convert("RGB")
+
+            img.thumbnail((width, height), Image.Resampling.LANCZOS)
+            w, h = img.size
+            tx = x + (width - w) // 2
+            ty = y_thumbnail + (height - h) // 2
+            grid.paste(img, (tx, ty))
+
+            # Add border
+            if BORDER_WIDTH > 0:
+                draw.rectangle(
+                    [
+                        (tx - BORDER_WIDTH, ty - BORDER_WIDTH),
+                        (tx + w + BORDER_WIDTH - 1, ty + h + BORDER_WIDTH - 1),
+                    ],
+                    outline="gray",
+                    width=BORDER_WIDTH,
+                )
+
+    return grid
+
+
+if __name__ == "__main__":
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/sandbox-scripts/xlsx/recalc.py

@@ -0,0 +1,178 @@
+#!/usr/bin/env python3
+"""
+Excel Formula Recalculation Script
+Recalculates all formulas in an Excel file using LibreOffice
+"""
+
+import json
+import sys
+import subprocess
+import os
+import platform
+from pathlib import Path
+from openpyxl import load_workbook
+
+
+def setup_libreoffice_macro():
+    """Setup LibreOffice macro for recalculation if not already configured"""
+    if platform.system() == 'Darwin':
+        macro_dir = os.path.expanduser('~/Library/Application Support/LibreOffice/4/user/basic/Standard')
+    else:
+        macro_dir = os.path.expanduser('~/.config/libreoffice/4/user/basic/Standard')
+    
+    macro_file = os.path.join(macro_dir, 'Module1.xba')
+    
+    if os.path.exists(macro_file):
+        with open(macro_file, 'r') as f:
+            if 'RecalculateAndSave' in f.read():
+                return True
+    
+    if not os.path.exists(macro_dir):
+        subprocess.run(['soffice', '--headless', '--terminate_after_init'], 
+                      capture_output=True, timeout=10)
+        os.makedirs(macro_dir, exist_ok=True)
+    
+    macro_content = '''<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE script:module PUBLIC "-//OpenOffice.org//DTD OfficeDocument 1.0//EN" "module.dtd">
+<script:module xmlns:script="http://openoffice.org/2000/script" script:name="Module1" script:language="StarBasic">
+    Sub RecalculateAndSave()
+      ThisComponent.calculateAll()
+      ThisComponent.store()
+      ThisComponent.close(True)
+    End Sub
+</script:module>'''
+    
+    try:
+        with open(macro_file, 'w') as f:
+            f.write(macro_content)
+        return True
+    except Exception:
+        return False
+
+
+def recalc(filename, timeout=30):
+    """
+    Recalculate formulas in Excel file and report any errors
+    
+    Args:
+        filename: Path to Excel file
+        timeout: Maximum time to wait for recalculation (seconds)
+    
+    Returns:
+        dict with error locations and counts
+    """
+    if not Path(filename).exists():
+        return {'error': f'File {filename} does not exist'}
+    
+    abs_path = str(Path(filename).absolute())
+    
+    if not setup_libreoffice_macro():
+        return {'error': 'Failed to setup LibreOffice macro'}
+    
+    cmd = [
+        'soffice', '--headless', '--norestore',
+        'vnd.sun.star.script:Standard.Module1.RecalculateAndSave?language=Basic&location=application',
+        abs_path
+    ]
+    
+    # Handle timeout command differences between Linux and macOS
+    if platform.system() != 'Windows':
+        timeout_cmd = 'timeout' if platform.system() == 'Linux' else None
+        if platform.system() == 'Darwin':
+            # Check if gtimeout is available on macOS
+            try:
+                subprocess.run(['gtimeout', '--version'], capture_output=True, timeout=1, check=False)
+                timeout_cmd = 'gtimeout'
+            except (FileNotFoundError, subprocess.TimeoutExpired):
+                pass
+        
+        if timeout_cmd:
+            cmd = [timeout_cmd, str(timeout)] + cmd
+    
+    result = subprocess.run(cmd, capture_output=True, text=True)
+    
+    if result.returncode != 0 and result.returncode != 124:  # 124 is timeout exit code
+        error_msg = result.stderr or 'Unknown error during recalculation'
+        if 'Module1' in error_msg or 'RecalculateAndSave' not in error_msg:
+            return {'error': 'LibreOffice macro not configured properly'}
+        else:
+            return {'error': error_msg}
+    
+    # Check for Excel errors in the recalculated file - scan ALL cells
+    try:
+        wb = load_workbook(filename, data_only=True)
+        
+        excel_errors = ['#VALUE!', '#DIV/0!', '#REF!', '#NAME?', '#NULL!', '#NUM!', '#N/A']
+        error_details = {err: [] for err in excel_errors}
+        total_errors = 0
+        
+        for sheet_name in wb.sheetnames:
+            ws = wb[sheet_name]
+            # Check ALL rows and columns - no limits
+            for row in ws.iter_rows():
+                for cell in row:
+                    if cell.value is not None and isinstance(cell.value, str):
+                        for err in excel_errors:
+                            if err in cell.value:
+                                location = f"{sheet_name}!{cell.coordinate}"
+                                error_details[err].append(location)
+                                total_errors += 1
+                                break
+        
+        wb.close()
+        
+        # Build result summary
+        result = {
+            'status': 'success' if total_errors == 0 else 'errors_found',
+            'total_errors': total_errors,
+            'error_summary': {}
+        }
+        
+        # Add non-empty error categories
+        for err_type, locations in error_details.items():
+            if locations:
+                result['error_summary'][err_type] = {
+                    'count': len(locations),
+                    'locations': locations[:20]  # Show up to 20 locations
+                }
+        
+        # Add formula count for context - also check ALL cells
+        wb_formulas = load_workbook(filename, data_only=False)
+        formula_count = 0
+        for sheet_name in wb_formulas.sheetnames:
+            ws = wb_formulas[sheet_name]
+            for row in ws.iter_rows():
+                for cell in row:
+                    if cell.value and isinstance(cell.value, str) and cell.value.startswith('='):
+                        formula_count += 1
+        wb_formulas.close()
+        
+        result['total_formulas'] = formula_count
+        
+        return result
+        
+    except Exception as e:
+        return {'error': str(e)}
+
+
+def main():
+    if len(sys.argv) < 2:
+        print("Usage: python recalc.py <excel_file> [timeout_seconds]")
+        print("\nRecalculates all formulas in an Excel file using LibreOffice")
+        print("\nReturns JSON with error details:")
+        print("  - status: 'success' or 'errors_found'")
+        print("  - total_errors: Total number of Excel errors found")
+        print("  - total_formulas: Number of formulas in the file")
+        print("  - error_summary: Breakdown by error type with locations")
+        print("    - #VALUE!, #DIV/0!, #REF!, #NAME?, #NULL!, #NUM!, #N/A")
+        sys.exit(1)
+    
+    filename = sys.argv[1]
+    timeout = int(sys.argv[2]) if len(sys.argv) > 2 else 30
+    
+    result = recalc(filename, timeout)
+    print(json.dumps(result, indent=2))
+
+
+if __name__ == '__main__':
+    main()

packages/twenty-server/src/engine/core-modules/code-interpreter/utils/get-mime-type.util.ts

@@ -0,0 +1,17 @@
+const MIME_TYPES: Record<string, string> = {
+  png: 'image/png',
+  jpg: 'image/jpeg',
+  jpeg: 'image/jpeg',
+  csv: 'text/csv',
+  xlsx: 'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet',
+  pptx: 'application/vnd.openxmlformats-officedocument.presentationml.presentation',
+  pdf: 'application/pdf',
+  json: 'application/json',
+  txt: 'text/plain',
+};
+
+export const getMimeType = (filename: string): string => {
+  const ext = filename.split('.').pop()?.toLowerCase();
+
+  return MIME_TYPES[ext ?? ''] ?? 'application/octet-stream';
+};