agentleFS
Sign inStart free
liush2yuxjtuliush2yuxjtu/html-artifacts/chatcut-production-mirror/AGENTS.md

html-artifacts / chatcut-production-mirror

AGENTS.md2 starsChanged 52 days agoApache-2.0View on GitHub

These instructions apply to all work under chatcut-production-mirror/. Keep the fast local and CI gates early: These checks are necessary but are not sufficient to call a product/UI change ready. After a Vercel Preview URL exists, use deterministic browser automation as the required post-preview acceptance gate. Preferred runners: Do not substitute a static fetch, build success, or verify:dist result for this browser check. If Deployment Protection is enabled, use Vercel's automation bypass support such as VERCEL_AUTOMATION_BYPASS_SECRET instead of relying on…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository6 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

10 rules. About 0 are common, 8 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Keep the fast local and CI gates early:

    line 7
  2. Distinctive:

    After a Vercel Preview URL exists, use deterministic browser automation as the required post-preview acceptance gate.

    line 19
  3. Very specific:

    Do not substitute a static fetch, build success, or verify:dist result for this browser check.

    line 26
  4. Very specific:

    If Deployment Protection is enabled, use Vercel's automation bypass support such as VERCEL_AUTOMATION_BYPASS_SECRET instead of relying on …

    line 28
  5. Distinctive:

    No framework error overlay or relevant browser console error appears during the checked flow.

    line 36
  6. Distinctive:

    When a stable visual baseline exists, run a screenshot diff and treat a material unexpected difference as an acceptance failure.

    line 38
  7. Distinctive:

    Use TinyFish only when human-like autonomous exploration adds value beyond the deterministic Playwright/agent-browser checks, or when the …

    line 46
  8. Distinctive:

    If TinyFish is unavailable, out of credits, or blocked by account capabilities, that does not block the PR when the required …

    line 48
  9. Distinctive:

    After the real Preview browser run, leave or update PR evidence containing:

    line 52
  10. Distinctive:

    Do not mark a product/UI PR ready for final human review only because local tests, GitHub Actions, or Vercel build checks are green.

    line 62

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

ChatCut agent instructions

These instructions apply to all work under chatcut-production-mirror/.

Shift Testing Left

Keep the fast local and CI gates early:

  1. npm test
  2. npm run build
  3. npm run verify:dist
  4. npm run ci:check for PR/local verification
  5. npm run ci:predeploy before Vercel publishes the output

These checks are necessary but are not sufficient to call a product/UI change ready.

Real Preview acceptance

After a Vercel Preview URL exists, use deterministic browser automation as the required post-preview acceptance gate.

Preferred runners:

  1. Playwright against the exact Vercel Preview URL.
  2. Vercel agent-browser when an agent is performing the verification interactively.

Do not substitute a static fetch, build success, or verify:dist result for this browser check.

If Deployment Protection is enabled, use Vercel's automation bypass support such as VERCEL_AUTOMATION_BYPASS_SECRET instead of relying on manually generated share links.

Verify at minimum:

  • / renders as the expected ChatCut production mirror without an obvious 404 or broken critical assets.
  • /intent.html is reachable and visibly renders.
  • Representative /features/... routes used by the change are reachable.
  • At least one affected homepage/demo interaction is exercised through the visible UI and reaches its expected visible result/state.
  • No framework error overlay or relevant browser console error appears during the checked flow.
  • Capture a screenshot and an interactive/DOM snapshot when the runner supports them.
  • When a stable visual baseline exists, run a screenshot diff and treat a material unexpected difference as an acceptance failure.

Any route, asset, interaction, console-visible failure, or access blocker discovered by the browser run is a real acceptance failure until resolved.

TinyFish policy

TinyFish is optional exploratory acceptance, not the default or mandatory PR gate.

Use TinyFish only when human-like autonomous exploration adds value beyond the deterministic Playwright/agent-browser checks, or when the user explicitly requests it. Do not make a PR depend on metered TinyFish availability when the same acceptance criteria can be expressed deterministically.

If TinyFish is unavailable, out of credits, or blocked by account capabilities, that does not block the PR when the required Playwright/agent-browser Preview acceptance passes.

PR evidence

After the real Preview browser run, leave or update PR evidence containing:

  • the exact Preview URL tested;
  • the browser runner used and run/artifact URL when available;
  • PASS / FAIL / BLOCKED;
  • routes and interaction(s) exercised;
  • screenshot/snapshot or visual-diff artifact references when available;
  • concrete failures or blockers;
  • queue, execution, and end-to-end timing separately when available.

Do not mark a product/UI PR ready for final human review only because local tests, GitHub Actions, or Vercel build checks are green. The real Preview browser acceptance must also pass.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free