| name | sync-chatgpt-deep-research-to-repo |
| description | End-to-end workflow to sync ChatGPT Project Deep Research PDFs into this git repo and refresh README indexes. Use when the user asks to download or sync Deep Research PDFs from ChatGPT via Playwright. |
Sync ChatGPT Deep Research PDFs into this repo
When to use
The user wants Playwright (typically user-playwright MCP) to pull Deep Research PDFs out of ChatGPT Projects into the repository and keep README indexes up to date.
Read first
- Repository agent guide:
AGENTS.md โ includes deep-research-download-log.json (last run + per-conversation PDF state).
- In order:
Execution order
- Session: Open ChatGPT via user-playwright MCP; complete login handoff if needed; list all Projects.
- Per project: Create or reuse
<repo>/<sanitized_project_name>/; enumerate threads with Deep Research; for each thread, export PDF through MCP (see PDF skill), save under that folder, then run scripts/sync_chatgpt_deep_research_pdfs.py record ... to update deep-research-download-log.json.
- README: Run
uv run python scripts/sync_chatgpt_deep_research_pdfs.py update-readmes (optionally --only <folder>) so each README.md PDF index matches files on disk.
- Root README: Update
../../README.md only if the project-folder layout or high-level description changed.
Done criteria
- Every new PDF appears under the correct project folder (saved via MCP, not the Python script).
deep-research-download-log.json conversations entries match what you did (record helper).
- No orphan table rows; every PDF listed in that folderโs README exists.
- Human was prompted if auth or UI blocked automation.