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 skills/anatomia-dev/anatomia/testing-standardsnpx skills add anatomia-dev/anatomia --skill testing-standardsgit clone --depth 1 https://github.com/anatomia-dev/anatomiaWhat 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 | $0.00034 | $0.00880 |
| Opus 5 | $0.00017 | $0.00440 |
| Sonnet 5 | $0.00007 | $0.00176 |
| Haiku 4.5 | $0.00003 | $0.00088 |
Grade A, and why
testing-standards 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 2d 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 — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Testing Standards
Detected
- Framework: Vitest (199 test files)
- Test command: pnpm run test -- --run
- Testing patterns: vitest
- Test location: dedicated test directory
Library Rules
- Always pass
--runflag when invoking Vitest in CI or non-interactive contexts. Vitest defaults to watch mode, which hangs pipelines waiting for input.
Rules
-
Test behavior, not implementation. Assert on what the code returns or produces — not which internal functions it calls. Tests should survive refactoring when behavior is unchanged.
-
Prefer real implementations over mocks. Mock only what you can't control: network calls, time, randomness. Every mock is a lie about how the system actually behaves.
-
Cover the error path, not just the happy path. For each feature test, write at least one test for invalid input, missing data, or service failure.
-
Assert on specific expected values from real inputs.
expect(status).toBe(200)notexpect(status).toBeDefined(). A test that passes regardless of whether the feature works catches nothing. Never write tautological tests —expect(true).toBe(true)proves nothing. The same applies to range matchers —toBe(2)nottoBeGreaterThan(0)when the test fixture has a known count. If you can't determine the specific expected value, read the contract'smatcher/valuefields before falling back to a weak assertion. When a test assertion matches the contract's matcher and value, the test is contract-aligned — not weak. -
Never weaken a test to make it pass. If a test fails, fix the code or fix the expectation — never broaden assertions or catch exceptions to force green.
-
When overriding EngineResult fields in tests, spread at each nesting level:
{ ...base, commands: { ...base.commands, test: 'vitest' } }. Never reconstruct the full object manually — the factory provides all required defaults and the type has 40+ fields. -
Never assert on source code content as a proxy for testing behavior — mock the trigger and assert on the output. Source-content assertions are acceptable as enforcement tests: verifying template content, security constraints, or structural invariants where the file content itself is what's being enforced.
-
Run the relevant test file after each change for fast feedback:
pnpm vitest run tests/{file}. Run the full suite before the build report to catch cross-file regressions. -
Scan integration tests run against the compiled
dist/index.js, not source. The pre-commit hook runs(cd packages/cli && pnpm run build)to keep the artifact fresh. If running tests manually outside the hook, run(cd packages/cli && pnpm run build)first — stale dist will give false passes on terminal output tests. -
Tests that create git repositories must force the branch name with
git init -b mainorgit branch -M mainafter first commit. CI runners have differentinit.defaultBranchsettings — never assume the default. This has caused CI failures 3 times. -
Use inline fixture data for scanner and parser tests — write files to temp directories at test time. Standalone manifest files with real package names (requirements.txt, package.json) trigger GitHub security advisory false positives.
-
Functions exported solely for test access (imported only by test files) are intentional — TypeScript has no
internalkeyword. Only flag exports with zero imports anywhere. -
When a test searches for expected output (find, findIndex, match on stdout or result arrays), assert the search succeeded — don't gate assertions behind a truthy check.
if (line) { expect(line).toContain(x); }silently passes when the output is missing. Writeexpect(line).toBeDefined()first, then assert the value. For optional detection results where absence is a valid test outcome, conditional checks are correct.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 38 lines · 34 tokens per session scan A 1f93113d780e
testing-standards is a skill published in the GitHub repository anatomia-dev/anatomia (32 stars, last pushed 29d ago), licensed MIT. It adds 34 tokens to every session and 880 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
agent-browser
Drive a real browser to inspect or interact with a web page or app — navigate, take screenshots, read console and network, fill simple forms — for verification tasks, not unattended automation.
rational-buddhism
当用户问某主张/玄学/灵修该不该信、想建立自己的验证标准、或寻找科学与精神的结合点时调用。 核心理念: 尝试一切, 亲自测试, 保持怀疑, 保留有用的, 舍弃没用的; 不可证伪的主张不进入基本真理; 冥想/接纳等内在技术可保留。 不适用于: 需要信仰安慰的临终/哀伤场景(尊重个体选择)。 Triggers: 该不该信/玄学/灵修/冥想/可证伪/验证/理性/rational buddhism/verify/belief.
requesting-code-review
Pre-commit review: security scan, quality gates, auto-fix.
workflow
Use when a task is too large for turn-by-turn orchestration and should run through the big-task workflow lane: system-wide changes, large migrations, repo-wide audits, high-confidence verification, or tasks explicitly asking to run a workflow. Claude Code uses native dynamic workflows; Codex, OpenCode, and Grok use…
verification-strategy
Thorough verification of completed work before declaring done.
gsd-verifier
Verifies phase goal achievement through goal-backward analysis. Checks codebase delivers what phase promised, not just that tasks completed.