agentleFS
lalulalilalulali/vespyr/.agents/skills/test/SKILL.md

test

Skill11 starsChanged 10 days agoMITView on GitHub

Run tests, analyze failures, enrich acceptance criteria, and produce structured QA reports — supports feature and full-cycle testing. Use when the user says "run the tests", "test this feature", "did I break anything", or invokes /test before merging or releasing.

186 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use agent-context-isolation, agentica-infrastructure and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 159 edited versions we read keep these.

  • 2

    Distinctive

    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:

    Follow the step sequence exactly.

    line 19
  2. Very specific:

    Coverage sweep (02o): every failure summary (feature and full-cycle) runs node .agents/scripts/req_to_impl_check.js and lists orphan ACs.

    line 23
  3. Distinctive:

    Before merging or releasing

    line 30
  4. Very specific:

    Test report produced at artifacts/output/05-execution/quality/test-report.md

    line 52

“Common” is measured: kept by at least 6 in 10 of the 159 edited versions of this file we could read. “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
SKILL.md

Test — Multi-Step QA Workflow

<!-- router-skill -->

Runs tests with depth — not just a pass/fail wrapper. Includes exploratory enrichment of acceptance criteria and separate feature vs. full-cycle test tracks.

Persona delegation

This skill delegates to @qa-engineer. The qa-engineer runs tests, analyzes failures, assesses regression risk, enriches acceptance criteria, and produces a test report. The skill file provides the workflow; @qa-engineer provides the analysis and multi-scenario testing depth. For eval-bearing tracks (judge prompts, scoring rubrics, benchmark suites, harness evals), consult @ml-ai-engineer on rubric design and production-distribution coverage before locking the report.

Harness adherence (non-negotiable)

  • Follow the step sequence exactly. Do NOT skip steps.
  • Step 01 (exploratory enrichment) is a hard gate — run it before any test execution.
  • Steps 02a and 02b are parallelizable but both must complete before step 03.
  • Each step file is a contract. Read it fully before executing.
  • Coverage sweep (02o): every failure summary (feature and full-cycle) runs node .agents/scripts/req_to_impl_check.js and lists orphan ACs. An unreferenced AC is a defect, not a detail — cite the sweep in the test report and qa-signoff (honestly labeled prose; the phase-advance refusal mechanism is a separately filed defect).
  • Metric validation (when artifacts/output/03-strategy/measurement-plan.md exists): verify the declared success metrics are actually instrumented before certifying; an uninstrumented success metric is a coverage gap, not a pass. The plan must conform first — run node .agents/scripts/analyze_data_check.js check --kind measurement-plan --file artifacts/output/03-strategy/measurement-plan.md --json. A plan whose metric rows carry no target number declares nothing to instrument, so the sweep has nothing to check against. If a dashboard is in scope, --kind dashboard-specs on artifacts/output/03-strategy/dashboard-specs.md tells you which panels claim to display each metric.

When to use

  • "Run the tests"
  • "Test this feature thoroughly"
  • "Did I break anything?"
  • Before merging or releasing

Step sequence

  1. Exploratory Enrichment → steps/step-01-exploratory-enrichment.md — review PRD/AC, storm assumptions, output missing scenarios 2a. Feature Testing → steps/step-02a-feature-test.md — unit/component tests in isolation 2b. Full-Cycle Testing → steps/step-02b-fullcycle-test.md — end-to-end integration workflows
  2. Criteria Backport → steps/step-03-criteria-backport.md — append discovered scenarios to PRD
  3. Completion → steps/step-04-completion.md — write test report, update status

Halt conditions

  • Step 01 (exploratory enrichment) has not run -> hard gate; testing unenriched acceptance criteria certifies the wrong contract
  • A failure cannot be reproduced locally -> halt and escalate to @tech-lead; an unreproducible failure is neither a pass nor a flake to wave through
  • A security finding is rated Critical or High -> halt; do not proceed past it
  • 2 QA review loops are exceeded -> halt and escalate to @tech-lead
  • The orphan-AC sweep reports an unreferenced acceptance criterion -> report it as a defect in the test report and the qa-signoff; an AC nothing implements is a coverage hole, not a detail

State machine integration

At start: run node .agents/scripts/orchestrator_state.js status At end: run node .agents/scripts/orchestrator_state.js complete --agent qa-engineer --artifact test-report.md

Done when

  • All steps completed
  • Test report produced at artifacts/output/05-execution/quality/test-report.md
  • Enriched acceptance criteria backported to user stories (step 03)
  • Minimum 3 newly discovered edge cases identified (step 01 verification)

Adversarial & Red-Team Data

AI-feature tests exercise the adversarial and red-team corpora under evals/suites/invariants/ — a green happy path with no adversarial case is an untested boundary.

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.