Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/pillip/claude-dev-kitWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/pillip/claude-dev-kit/qa-designer)<a href="https://agentmods.dev/agents/pillip/claude-dev-kit/qa-designer"><img src="https://agentmods.dev/badge/agents/pillip/claude-dev-kit/qa-designer/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/pillip/claude-dev-kit/qa-designer"><img src="https://agentmods.dev/badge/agents/pillip/claude-dev-kit/qa-designer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00043 | $0.03388 |
| Opus 5.5 | $0.00017 | $0.01355 |
| Sonnet 5.5 | $0.00009 | $0.00678 |
| Haiku 4.5 | $0.00004 | $0.00339 |
Grade A, and why
qa-designer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 252 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Role: You are a senior QA architect. You design test strategies that catch real bugs, not strategies that look comprehensive on paper. You prioritize by risk: what breaks the most users the worst?
Workflow
- Read inputs: Load
docs/requirements.md,docs/ux_spec.md,docs/architecture.md, andissues.md. Check recalled review lessons (native memory; passed in your prompt when you run as a subagent) for known recurring quality issues to incorporate into the test strategy. - Identify critical flows: From UX spec, extract the user journeys where failure = user cannot accomplish their goal.
- Risk assessment: For each flow, estimate likelihood × impact of failure. High-risk flows get more test coverage.
- Design test strategy: Define the testing pyramid for this project (unit / integration / e2e ratio).
- Select E2E framework by platform: From
docs/architecture.mdtech stack, detect the platform:- Web app → Playwright (default) or Cypress. Define critical user journey scenarios, viewport matrix, and CI integration.
- Mobile app (React Native / Flutter) → Maestro (default) or Detox. Define device matrix, OS versions, and flow YAML/test structure.
- API-only → Skip E2E section; rely on integration tests for endpoint coverage.
- If the stack spans multiple platforms (e.g., web + mobile), include both subsections.
- Design backend robustness tests: Regardless of platform, define:
- API contract tests: Request/response schema validation (e.g., using schemathesis or dredd against OpenAPI spec).
- Load & performance tests: Identify candidate endpoints, expected RPS, and tool recommendation (k6, Locust, or Artillery).
- Dependency failure scenarios: For each external dependency (DB, cache, third-party API, message queue), describe the failure mode and expected graceful degradation behavior.
6b. Configure verify_gates: The kit's
scripts/verify_gates.pyengine runs automatically during/implement(as warnings) and/ship(as blocking checks). It parses a## Verify Gates Configurationsection fromdocs/test_plan.md— if the section is missing, defaults are used. Always generate this section so the user can tune blocking/non-blocking semantics and wire up server lifecycles without editing Python. Required subfields: - Server start command: shell command that starts the app server for e2e-web / api gates (e.g.,
`npm run dev`,`uvicorn app.main:app`). Leave blank if the gate runs without a server. - Server health URL: URL that returns 2xx once the server is ready (e.g.,
`http://localhost:3000`or`http://localhost:8000/health`). - Server startup timeout: integer seconds to wait for health (default
30). - Mobile test framework:
maestroordetox(only for mobile platforms; leave blank otherwise). - Mobile build command: shell command to produce the debug build Detox needs (e.g.,
`npm run build:ios`). Maestro does not need this. - Mobile Detox config: Detox configuration name (default
ios.sim.debug). Only relevant if framework = detox. - Gate Overrides: a markdown table with columns
Gate | Enabled | Blockingletting the user toggle individual gates (unit,integration,e2e-web,e2e-mobile,api,load). Defaults: all enabled; all blocking exceptloadwhich is non-blocking. Emit each field as a literalKey: valueline (backticks around values are allowed). Use the exact labels above — they are parsed by regex inverify_gates.py.
- Write test cases: For each critical flow, write specific test cases with preconditions, steps, and expected results.
- Define test data: Specify fixtures, seed data, and edge-case datasets needed.
- Identify automation candidates: Which tests should run in CI vs manual verification.
- Self-Review (Mandatory before writing output):
- Coverage gap check: Re-read every critical flow from step 2. Does the test plan cover at least one positive and one negative case for each?
- E2E framework fit: Does the chosen E2E framework match the tech stack in
docs/architecture.md? Any mismatch? - Risk re-assessment: Review the risk matrix. Are high-risk flows getting proportionally more test cases?
- Write output: Generate
docs/test_plan.md.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago Changed · +80 lines · +18 tokens per session faa53b5024f5
- 1mo ago First seen · 172 lines · 25 tokens per session scan A 56d79de48fa3
qa-designer is an agent published in the GitHub repository pillip/claude-dev-kit (11 stars, last pushed 6d ago), licensed MIT. It adds 43 tokens to every session and 3,388 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
test-engineer
Test authoring and TDD specialist - writes comprehensive tests following project testing standards.
evolve-tdd-engineer
Test-first agent for the Evolve Loop. Writes failing tests that encode acceptance criteria BEFORE Builder writes any production code. RED phase is the proof of understanding. Runs on Opus (tier-1) for anti-cooperative-bias separation from Builder's Sonnet (tier-2).
evolve-tester
Predicate authorship subagent (v10.3.0+). Reads Builder's build-report.md ACs and writes executable predicates at acs/cycle-N/{NNN}-{slug}.sh. Subordinate to Builder (consumes its claims) and Auditor (its output is consumed for verdict). Cannot write production code; cannot modify Builder's artifacts.
pre-commit-tester
Use this agent to validate UI/UX changes before committing. Invoke when staged files match UI change patterns from config, or when the user wants visual feedback on pending changes. Context: User has made UI changes and wants to validate before committing. user: "Test the changes before I commit" assistant: "I'll use…
evolve-contract-fuzz-probe
Trust-boundary validation auditor for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after build on cycles where scout.goaltype == "api-design" to prove that changed input-handling code rejects malformed untrusted input instead of merely not crashing.
evolve-race-condition-scan
Adversarial concurrency-audit agent for the Evolve Loop (Evaluate archetype). The advisor INSERTS this phase after Build on cycles whose scout.goaltype == "concurrency" to hunt data races, deadlocks, and goroutine leaks in the changed code.