bugfix-workflow
Use when the user reports a concrete bug in an existing application and asks for a code fix. Reproduce the bug, write or run a verifier that fails, change the smallest amount of code that makes the verifier pass, re-run the verifier, and explain the change. Do not use for: feature requests, code review, dependency upgrades, architecture changes, or "make this faster" requests with no failing test.
- License 许可
- MIT
- 最近更新
- 2026-07-28
- Stars
- ★ 3
- Forks
- 0
01 前言 Front Matter
- name
- bugfix-workflow
- description
- Use when the user reports a concrete bug in an existing application and asks for a code fix. Reproduce the bug, write or run a verifier that fails, change the smallest amount of code that makes the verifier pass, re-run the verifier, and explain the change. Do not use for: feature requests, code review, dependency upgrades, architecture changes, or "make this faster" requests with no failing test.
03 SKILL.md 正文
bugfix-workflow
Promise: a failing repro turns into a passing repro through a minimal, explained code change.
Steps
- Read the bug report. Identify the failing observable behaviour.
- Locate or write a verifier (test, script, manual repro) that exhibits the bug. Run it and confirm it fails.
- Change the smallest amount of code that plausibly fixes the bug.
- Re-run the verifier. If it still fails, iterate. If it passes, stop.
- Explain the change in 2–3 sentences: what was wrong, what you changed, why it fixes it.
Anti-patterns
- Do not modify files unrelated to the bug.
- Do not bypass tests with
--no-verifyor by deleting assertions. - Do not push to a protected branch (e.g.
main) without explicit instruction. - Do not read or print secrets (
.env,credentials.*, files undersecrets/).
04 同仓库更多 Skills
agent-eval-skills
Evaluate an existing agent skill with Promptfoo-native behavior, routing, or role-play tests. Use when a skill needs executable evidence that its product promise works. Do not use to author the domain skill itself.
verifier-skill
Use when checking verifier wiring. Do not use for live agent behavior.
runtime-fields
Use when validating Agent Skill Evals runtime test fields. Do not use for real agent runs.
missing-frontmatter
暂无简介
fixtureless-check
Use when validating repository-level checks that do not need a copied fixture. Do not use for agent behavior.
broken-routing
Helps with GitHub workflows.