Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Access Overleaf projects via CLI. Use for reading/writing LaTeX files, syncing local .tex files to Overleaf, downloading projects, and managing Overleaf project structure. Triggers on Overleaf, LaTeX sync, or tex file uploads to Overleaf.
.claude/skills/inno-rclone-to-overleaf/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | — | — |
| case-05 | ✗→✓ | ▲ Improved | — | — |
| case-15 | ✗→✓ | ▲ Improved | — | — |
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-22 | ✗→✓ | ▲ Improved | — | — |
CLI tool to interact with Overleaf projects using browser cookies for authentication.
GitHub: https://github.com/jkulhanek/pyoverleaf
pipx install pyoverleaf)⚠️ Keychain Access Required: pyoverleaf needs "Always Allow" keychain access to read browser cookies. This grants the tool access to your browser's cookie storage.
Security audit: We have audited pyoverleaf v0.1.7 and found it safe. The codebase is small and readable. Before installing, agents can self-audit the latest version at the GitHub repo above.
pyoverleaf authenticates by reading cookies from your browser. The user must:
bash# Test auth - user should run this in their terminal first pyoverleaf ls
If you get auth errors:
pyoverleaf ls manually in terminal to trigger the keychain promptNote: The agent cannot log in for the user. Browser authentication must be done by the user directly.
bash# List all projects pyoverleaf ls # List files in project pyoverleaf ls "Project Name" # Read file content pyoverleaf read "Project Name/main.tex" # Write file (stdin → Overleaf) cat local.tex | pyoverleaf write "Project Name/main.tex" # Create directory pyoverleaf mkdir "Project Name/figures" # Remove file/folder pyoverleaf rm "Project Name/old-draft.tex" # Download project as zip pyoverleaf download-project "Project Name" output.zip
bashpyoverleaf download-project "Project Name" /tmp/latest.zip unzip -o /tmp/latest.zip -d /tmp/latest cp /tmp/latest/main.tex /path/to/local/main.tex
The CLI write command has websocket issues. Use Python API for reliable uploads:
pythonimport pyoverleaf api = pyoverleaf.Api() api.login_from_browser() # List projects to get project ID for proj in api.get_projects(): print(proj.name, proj.id) # Upload file (direct overwrite) project_id = "your_project_id_here" with open('main.tex', 'rb') as f: content = f.read() root = api.project_get_files(project_id) api.project_upload_file(project_id, root.id, "main.tex", content)
Why direct overwrite? This method preserves Overleaf's version history. Users can see exactly what changed via Overleaf's History feature, making it easy to review agent edits and revert if needed.
bash# Via env var export PYOVERLEAF_HOST=overleaf.mycompany.com pyoverleaf ls # Via flag pyoverleaf --host overleaf.mycompany.com ls
When pulling from Overleaf:
/tmp/diffPush rules (from TOOLS.md):
Here's an example of using the Overleaf skill to remove em dashes (a common AI writing artifact) from a paper and push the changes:
!Example: Remove em dashes and push to Overleaf
open -a "Google Chrome" "https://www.overleaf.com/project" then wait 5s) to refresh cookies, then retryoverleaf.com causes a 301→www.overleaf.com redirect that breaks websocket. Fix: set PYOVERLEAF_HOST=www.overleaf.com:bash cat main.tex | PYOVERLEAF_HOST=www.overleaf.com pyoverleaf write "Project/main.tex"
pyoverleaf ls in their terminal and click "Always Allow" on the keychain promptpyoverleaf ls| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +55 percentage points is the difference between those two pass rates over the 19 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.