Overcut named in the 2026 Gartner® Innovation Insight: AI Software Factories reportRead more

AGENTS.md that stays current

Keeps every repo's AGENTS.md in sync with reality — Overcut analyzes your structure and patterns on a schedule and opens a PR with the doc changes that actually matter.

Try it now
Auto Update AGENTS.md — example generated output

Self-maintaining agent instructions

On a schedule, Overcut clones the repo, finds or starts the right branch, analyzes structure and conventions, regenerates AGENTS.md, filters out the noise, and commits a clean PR only when something real changed.

Auto Update AGENTS.md — how the workflow runs, trigger to result

The edge cases, covered

A significance check discards stylistic churn and the workflow reuses open drafts, so you get meaningful updates instead of a stream of trivial PRs.

Significance filter

Reviews the diff and discards purely stylistic or formatting changes, reverting the file so you never get a noise PR.

Reuses open drafts

Finds an existing AGENTS.md PR and builds on its branch instead of opening a duplicate, keeping updates incremental.

Pattern recognition

Scans workflows, READMEs, and code structure to capture new conventions and best practices automatically.

Skips when nothing changed

If analysis finds no meaningful updates, it exits cleanly without committing or creating a pull request.

Adapts to any repo

Tailors the generated sections to whatever the repository contains and preserves your existing structure and style when updating.

Drops into any repository

Works with any repo on your git host — reading workflows, READMEs, and code to document the patterns it finds, then raising a pull request.

GitHub
GitLab
Overcut

Running from day one

Runs automatically on a daily cron schedule you can tune to any interval, or fire it on demand with /update-agents-md to regenerate immediately.

Auto Update AGENTS.md — a run timeline from a real execution

Trusted by industry leaders worldwide

Dvir Cooper

“In just four months, Overcut has saved us countless hours and quickly become an indispensable tool that developers actively request for new projects.”

Dvir Cooper,R&D Group Lead, Zesty

Read the full story

100+

hours saved in the first four months

9/10

new projects start on overcut

Your AI factory starts with one command

Book a 30-min demo call
Claude logoOpenAI logoCursor logoGitHub Copilot logo
~/acme-app
$ npx overcut init
✓ Signed in
✓ Current workspace: Acme
✓ Connected: Claude Code
> Set up Overcut for this repo.
Inspecting repo: GitHub remote, pnpm, vitest, GitHub Actions
Proposed workflows:
1. Code review
2. Fix CI
3. Auto docs update on merge
Draft run of "Code review" passed.
Publish and activate triggers? [y/N]