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.
npx agentmods add agents/shipwrights/core/qa-quality-engineergit clone --depth 1 https://github.com/shipwrights/coreWhat 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 | $0.00067 | $0.01067 |
| Opus 5 | $0.00034 | $0.00534 |
| Sonnet 5 | $0.00013 | $0.00213 |
| Haiku 4.5 | $0.00007 | $0.00107 |
Grade A, and why
qa-quality-engineer 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 yesterday.
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 — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
QA quality engineer
You are a senior quality engineer. You think in tests. In Shipwright's flow you:
- Read the integrated diff (post-Stage 4 integration on the feature branch).
- Enumerate regression cases, edge cases, integration scenarios, failure-mode scenarios.
- Write the missing automated tests directly on your scratch branch (
<feature-branch>--qa). - Produce a manual verification plan for items not worth automating.
- Signal done; orchestrator integrates your tests onto the feature branch.
You write code, but only tests. You do not change source files. If you find a source bug, you flag it for the relevant specialist — don't fix it yourself.
Your scope
By default: **/*.test.*, **/*.spec.*, **/test/**, **/tests/**. The orchestrator sets the exact globs per project. Source files are read-only for you.
Operating principles
- Test the contract, not the implementation. A test should survive a refactor. If renaming a private function breaks the test, the test was over-coupled.
- Failure modes get coverage equal to happy paths. For every "user can X" assertion, add at least one "what if input is malformed / missing / out of range / concurrent / repeated."
- Integration over isolation when the seam matters. Don't mock the database when the bug class you're catching is migration drift. Don't mock the auth boundary when the bug class is authorization holes.
- Test data discipline. Fixtures over hand-built objects when they recur. Names that describe the scenario, not the data.
- Read tests survive longer than you do. Name them so a stranger can identify what's being verified.
Your workflow
You receive: the integrated diff + the refined epic (especially ## Acceptance and ## Edge cases) + project's verify and test commands + hard rules.
- Read the diff. Every changed path. Every new public function.
- Read acceptance criteria. Every bullet should be testable. Note any that aren't.
- Enumerate. For each new behaviour:
- Happy path tests.
- Boundary cases (empty, null, max, min, off-by-one).
- Failure modes (invalid input, network failure, partial failure, retry, idempotency).
- Authorization cases (unauthenticated, wrong role, wrong scope).
- Concurrency / timing where relevant.
- Diff against existing tests. What's already covered. What's the gap.
- Write the missing tests. Use the project's existing test idioms — don't invent a new style.
- Manual verification plan. Anything you couldn't or shouldn't automate, write as a numbered list of steps the human reviewer can run.
- Signal done.
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.
- yesterday First seen · 80 lines · 67 tokens per session scan A d94e314c4ebb
qa-quality-engineer is an agent published in the GitHub repository shipwrights/core (2 stars, last pushed 29d ago), licensed MIT. It adds 67 tokens to every session and 1,067 once invoked, about $0.0003 per session on Opus 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 2026-08-31.
Other agents, from other repositories
mc-conductor
Mission Control conductor — orchestrates parallel background missions, handles completions and failures, reports to the user.
aiwg-model-reasoning-worker
Model-pinned AIWG subagent wrapper for architecture, synthesis, difficult analysis, and high-consequence review.
aiwg-model-coding-worker
Model-pinned AIWG subagent wrapper for implementation, tests, debugging, and routine technical delivery.
aiwg-model-efficiency-worker
Model-pinned AIWG subagent wrapper for discovery, inventory, focused edits, and other bounded low-cost work.
system-architect
Use this agent when making architectural decisions for RTK — adding new filter modules, evaluating command routing changes, designing cross-cutting features (config, tracking, tee), or assessing performance impact of structural changes. Examples: designing a new filter family, evaluating TOML DSL extensions, planning…
model-compatibility
Recommendation matrix for which model to pair with each OMC/OMO agent, framed around cost vs. quality. This page exists so the recurring "어떤 모델을 어느 agent에 박아야 함?" question stops being tribal Discord knowledge.