Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/WangChangxin0809/repo-agent-harnessnpx agentmods add skills/wangchangxin0809/repo-agent-harness/writing-checksWrote 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/skills/wangchangxin0809/repo-agent-harness/writing-checks)<a href="https://agentmods.dev/skills/wangchangxin0809/repo-agent-harness/writing-checks"><img src="https://agentmods.dev/badge/skills/wangchangxin0809/repo-agent-harness/writing-checks/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/skills/wangchangxin0809/repo-agent-harness/writing-checks"><img src="https://agentmods.dev/badge/skills/wangchangxin0809/repo-agent-harness/writing-checks.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 | $0.00105 | $0.02841 |
| Opus 5 | $0.00053 | $0.01421 |
| Sonnet 5 | $0.00021 | $0.00568 |
| Haiku 4.5 | $0.00011 | $0.00284 |
Grade A, and why
writing-checks 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 11d 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 — 266 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Guards and gates
Governs: shared/scripts/guards/, shared/scripts/gates/
Two mechanisms, one discipline. A guard reads one proposed action before it runs and may block it. A gate reads the worktree at CI time and may fail the build. Everything below applies to both, because the failure modes are the same.
| Guard | Gate | |
|---|---|---|
| Sees | one tool call, as JSON on stdin | the whole tree |
| Cost | every matching call | once per run |
| Use when | the action is irreversible | the state is detectable |
| Failure | exit 2, stderr goes to the model | non-zero, output goes to a human |
A guard is a speed bump, not a boundary
Say this out loud before writing one, because the rest of this skill reads like a promise it cannot keep. A guard matches the text of a proposed command:
git push origin main | tail -5 BLOCKED
B=push; git $B origin main | tail -5 allowed
It also fails open on purpose — a broken guard must not become an unbypassable wall — so its coverage is best-effort in two independent directions. That is the right trade for what a guard is actually for: catching the thing you were about to do out of habit, and explaining why, at the moment you were doing it.
It is the wrong trade for anything adversarial, and for anything where a single miss is unacceptable. Those get three layers, in this order:
| Mechanism | Why it is stronger | |
|---|---|---|
| 1 | permissions.deny in .claude/settings.json |
Evaluated by the harness, not by a regex you maintain |
| 2 | Server-side branch protection, required CI | Survives the laptop, the config, and the plugin entirely |
| 3 | A guard here | Explains why, in prose, at the moment of the attempt |
Prefer a deny rule for anything a deny rule can express:
"deny": ["Bash(git push --force:*)", "Bash(git push -f:*)"]
Write a guard when the rule is conditional in a way a deny pattern cannot state
— no_protected_branch_push.py is the worked example: the rule depends on the
refspec, so Bash(git push:*) would block the branch you are allowed to push.
And write a guard when the prose is the product, which is more often than it
sounds. The paragraph on stderr is the highest-value text in the repository.
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.
- 11d ago First seen · 266 lines · 105 tokens per session scan A 9321db6bc0d3
writing-checks is a skill published in the GitHub repository WangChangxin0809/repo-agent-harness (2 stars, last pushed 12d ago), licensed MIT. It adds 105 tokens to every session and 2,841 once invoked, about $0.0005 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 skills, from other repositories
guards-setup
A setup guide for installing automatic code-quality checks, such as linting, type checking, dependency-cycle checks, and optional CI checks.
verify
The profile-parameterized static-validation + test executor (/foundry:verify, SDLC steps 7 & 8). Resolves the ACTIVE stack profile from .foundry/stack-profile.lock (via the merged stack-profile loader's resolvelock) and dispatches the profile-declared staticvalidation (format/lint/typecheck/build) + testrecipe…
docs-verify
Validates documentation quality and freshness — checks for broken links, stale content, llms.txt sync, image issues, heading hierarchy, and badge URLs. Runs locally or in CI. Use to catch documentation decay before it reaches users.
skybreaker
Turn expectations into enforceable law: lint, types, contracts, CI.
playwright-testing
E2E testing with Playwright - Page Objects, cross-browser, CI/CD.
tdd-cycle
Execute full TDD red-green-refactor cycle with validation gates. Use when saying "TDD cycle", "test-driven development", or "full TDD workflow".