kimbell-skills

Plugin kimbell-skills from marketplace kimbell in TzolkinB/skills. Commands read from the repo's manifest — 3 days ago.

Skills (15)

• noop-skill

  Seeded fixture for lint --self-test — do not ship

• audit-orchestrator

  Route a suspicious *passing* test to the tool that can actually prove it — unit JS/TS → Tautest/StrykerJS, app-driven Playwright/Cypress → audit-test with the reachability guard — and emit a provenance-labelled verdict, never a gate.

• audit-test

  Audit a passing test for false confidence — would it still pass if the code it covers broke? — by running a targeted mutation and reporting what actually happened, not by reasoning alone

• bug-report

  Convert a messy failure description into a structured, actionable bug report

• contract-guard

  Consumer-side contract check for the stranded frontend team Pact can't serve — tiered, cheapest-first — detect existing response validation, else recommend/scaffold client-side validation (and, if a published spec is available, also recommend test-boundary validation with its per-operation drift-cov

• coverage-review

  Review test coverage and flag untested paths, edge cases, and assertion gaps — reads real coverage instrumentation when present, falls back to static analysis

• debug-test

  Automatically diagnose a failing Playwright test — reads files directly, applies QA heuristics, routes to the Playwright healer or diagnosing-bugs, and classifies what the healer changed before calling it fixed, flagging 🔁 when the same test keeps getting healed for the same reason (read from git h

• e2e-impact

  Map a working/PR diff to the Playwright/Cypress E2E specs it plausibly hits — via test-import, route, and selector/test-id signals — with a confidence per spec and an honest run-all fallback when a change can't be traced. Emits a source→spec relevance map.

• gate

  The Gate stage (stage 7): binds a PR's existing E2E results (Playwright and/or Cypress) plus an audit-test verdict into one evidence bundle, and derives an advisory ship/canary/hold decision by worst-wins. Ship requires the E2E run to clear an executed-floor and the audit-test verdict to clear an ex

• prune-tests

  Review a test file or suite for low-value, redundant, over-mocked, or stale tests and propose a conservative prune / merge / rewrite plan — the subtractive counterpart to coverage-review

• qa-compass

  Router — describe your QA situation (and, if you have one, your stack) and get pointed at the best QA-AI tool for it, one of these skills or an external one, across the seven-stage map — with why it fits, the exact way to run it, its evidence label, and where it sits in the flow. The runnable front

• qa-pass

  Run a QA judgment pass over a feature branch — test plan gaps, coverage, assertion quality, false-confidence audit, regression risks

• qa-review

  Review code from a QA/testability angle — is this even testable?

• test-plan

  Generate a structured QA test plan from a feature description or ticket

• threat-model

  Threat model a change — what actually breaks in production if it's wrong, and how you'd find out — independent of whether it's tested

Similar plugins