AGENT SKILL

mr-impact

WAY 11 — run only the tests a merge request / pull request puts at risk and report on it (GitLab MR, GitHub PR, Azure DevOps PR, Bitbucket PR). Use when the user says "test mr", "test pr", "mr impact", "pr impact", "what does !87 break", "check this pr" with an MR/PR number or URL.

License 许可
MIT
最近更新
2026-11-06
Stars
★ 5
Forks
4

01 前言 Front Matter

name
mr-impact
description
WAY 11 — run only the tests a merge request / pull request puts at risk and report on it (GitLab MR, GitHub PR, Azure DevOps PR, Bitbucket PR). Use when the user says "test mr", "test pr", "mr impact", "pr impact", "what does !87 break", "check this pr" with an MR/PR number or URL.

03 SKILL.md 正文

WAY 11 — MR/PR Test Impact

Run CLAUDE.md Step 0 first. For a change request, the right target is usually its review/preview app (--env feature/<branch>). Default to staging only if the user confirms the change is deployed there.

Phase 1: Select

Run mr-test-impact with the MR/PR number (or URL), QA_CODE_HOST, and that provider's non-secret settings. It returns the selected tests (with reasons and confidence), the folder scope filters, and the changed areas no test covers. The diff stays in its context.

If nothing is selected and broad_change is false, say so and go straight to Phase 3 with the uncovered list.

Phase 2: Run

Run test-execution-agent per project (e2e, api) with the returned scope_filters. Order everything by priority (P0 first), then confidence: a failing P0/P1 test is what blocks the merge.

  • failure_kind: locator → test-healer (propose mode). A label change in this MR/PR is the usual cause, so show the heal next to the change itself
  • Don't file bugs automatically. The code isn't merged yet, so failures go to the author first. File bugs only if the user asks

Phase 3: Report on the MR/PR (change.comment, per the tracker skill)

Post one comment (use the provider's ref format: !87 on GitLab/Azure, #87 on GitHub/Bitbucket). Rows are sorted by priority, then result (failures first):

## 🧪 QA impact — <ref>

**Environment:** <url> · **Run:** <RUN_ID>
**Selected:** N tests (H high / M medium / L low confidence) · ✅ P passed · ❌ F failed · ⚠️ B blocked
**Blocking:** ❌ K failed P0/P1 tests (or "none")

| Test | Priority | Why selected | Result |
|---|---|---|---|
| Test#03 — Checkout applies discount | P0 | `POST /v1/orders` changed (src/orders/controller.ts) | ❌ step 3: total not updated |

**Tests needing a heal (label changed in this change):** …
**Changed areas with no test:** `/settings/billing` (src/pages/settings/billing.tsx). Run `write it` to cover it.

---
🤖 _Generated by DevOps QA Agent — review before acting on this._

Also save it to ./qa-artifacts/sessions/change-<ref>-<YYYY-MM-DD-HH-MM>.md.

04 同仓库更多 Skills

GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

write-tests

WAY 4 — write test cases only, no execution. Use when the user says "write it", "write tests", "generate test cases" — an approved test plan (`Plan:`), or requirements (text or a tracker issue) planned and approved first, optional App/Spec/MR or PR → exactly the planned test files.

★ 5
GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

verify-fix

WAY 9 — verify that a bug is fixed by running its regression test. Use when the user says "verify it", "verify fix", "is #142 fixed", "check the fix" with one or more bug refs (#142, SHOP-142), or "verify all" for every open bug that has a regression test.

★ 5
GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

user-story-format

Reference only — the single definition of this project's user-story body (story, Gherkin ACs, Missing Information, DoD) and its AC/scenario rules. Read by user-story-enhancer (WAY 6) and benchmark-gap-story-writer (WAY 8) so stories they write stay identical in shape and WAY 6 can re-score WAY 8's output. Not a workflow on its own.

★ 5
GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

tracker

Provider-neutral contract for every issue-tracker and code-host action (create/update/comment/close issues, attach files, comment on MRs/PRs, read CI runs and logs) plus label conventions and failure handling. Every WAY calls these operations; the matching provider-* skill says how to perform them on GitLab, GitHub, Jira, Azure DevOps, or Bitbucket.

★ 5
GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

traceability

WAY 14 — requirements traceability matrix for a user story (each acceptance criterion) or a test folder written from plain text (each R-NN requirement) → the tests covering it → their latest result → open bugs. Use when the user says "trace it", "traceability", "is #148 covered", "coverage of SHOP-148", "trace it <folder>".

★ 5
GhassenAZZOUZ
GhassenAZZOUZ/claude-qa-agent

test-plan

WAY 19 — design a test plan before any test code. Use when the user says "plan it", "plan tests", "test plan", "design tests" with requirements (text or a tracker issue), optional App/Spec/MR or PR. Output is the test plan (scope, levels, one row per Test#<nn> with category, tags, covers, priority, coverage, entry/exit criteria, risks), reviewed by `review-plan` (WAY 20), filed for approval; `write it Plan:` / `e2e it Plan:` then write exactly that plan. Also holds the plan format every other WAY reads.

★ 5