agentsclimarketplace

Repro first

Skill Rookage/agent-discipline/skills/repro-first

Practical bilingual skills that make AI coding agents more disciplined, testable, and useful in real engineering work.

Install
npx -y skills add Rookage/agent-discipline --skill repro-first

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when debugging bugs, failures, broken behavior, flaky tests, regressions, crashes, or performance issues. Build a reproducible feedback loop before hypothesizing or fixing; useful when the user says debug, diagnose, failing, broken, cannot use, slow, or error.

SKILL.md

1.6 KB, as published. Nobody here has run it

Repro First

中文核心:先复现,再修。不要猜。

Workflow

  1. Build a feedback loop.

    • Prefer failing test, curl/HTTP script, CLI repro, Playwright script, trace replay, or minimal harness.
    • The loop must produce a clear pass/fail signal.
  2. Reproduce the user-visible failure.

    • Confirm the symptom matches what the user reported.
    • Capture exact error, output, timing, request, DOM state, or log.
  3. Create 3-5 ranked hypotheses.

    • Each hypothesis must be falsifiable.
    • Format: if X is the cause, then changing or observing Y will prove or disprove it.
  4. Instrument narrowly.

    • One variable at a time.
    • Use targeted logs, breakpoints, assertions, or probes.
    • Prefix temporary logs with [DEBUG-repro-first] and remove them before finishing.
  5. Fix with regression protection.

    • If a correct seam exists, write the failing regression test before the fix.
    • If no correct seam exists, document that the architecture blocks reliable regression coverage.
  6. Verify.

    • Re-run original repro.
    • Re-run relevant tests.
    • Remove temporary artifacts.

Communication

  • State evidence, not vibes.
  • Do not ask the user to run commands the agent can run.
  • Ask for user input only when the missing artifact is outside agent reach.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.