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 skills add vasuag09/harness-claude --skill fixgit clone --depth 1 https://github.com/vasuag09/harness-claudeWrote 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/vasuag09/harness-claude/fix)<a href="https://agentmods.dev/skills/vasuag09/harness-claude/fix"><img src="https://agentmods.dev/badge/skills/vasuag09/harness-claude/fix.svg" alt="Measured on agentmods" 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 | $0.00138 | $0.01745 |
| Opus 5 | $0.00069 | $0.00873 |
| Sonnet 5 | $0.00028 | $0.00349 |
| Haiku 4.5 | $0.00014 | $0.00175 |
Grade A, and why
fix 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 8d 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 — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/fix — reproduce first, fix minimally, ride the gates
Goal: take something that is already broken and fix it with discipline — capture the bug as a
failing test before touching the fix, scope the change to the cause, and let it flow through the
harness's existing review/verify/ship gates. The pipeline's front door (/spec → … → /ship) is
greenfield-shaped — it assumes you're building. /harness-claude:fix is the parallel entry for bugs: it
does the bug-specific work, then hands off rather than re-implementing the gates.
Opt-in. Invoked explicitly; nothing auto-routes a bug here. Git boundary:
/harness-claude:fixnever commits or pushes — branch creation for non-trivial work is expected (claude/fix-<slug>, seerules/git.md); the fix is committed (if at all) only through/harness-claude:shipwhen you ask. Reuse, don't rebuild: the regression test goes throughharness-claude:tdd-guide; review/verify/ship are the existing skills, not duplicated here.
When /harness-claude:fix, when /harness-claude:build-fix
/harness-claude:fixis for behavioral / logic bugs — wrong output, a crash on some input, a regression, a reported defect. It is reproduce-as-a-failing-test-first, then root-cause, then minimal fix./harness-claude:build-fixis for build / type / compile errors — get the build green with a minimal diff. It is not a competitor: while fixing a behavioral bug you may call/harness-claude:build-fixto keep the build green mid-change. Use/harness-claude:build-fixalone when nothing is behaviorally wrong and you just need the compiler/types happy.
How it works
reproduce (RED test) → root cause → minimal fix-plan → hand off to /implement → /verify → /ship
Do this
- Branch, then reproduce — capture the bug RED. Apply branch-at-first-write
(
rules/git.md) before the test file lands: not a git repo → skip silently; already on a non-default branch → stay there; on the default branch with non-trivial work → createclaude/fix-<slug>first (never commit or push — that stays gated behind your explicit ask). Then, before writing any fix, produce a failing test that exercises the reported behavior and fails for the right reason. Run it; confirm it's RED. If a reliable automated repro genuinely isn't possible (e.g. needs real hardware, an external outage), say why and record concrete manual repro steps instead — never skip the repro step silently. - Find the root cause, not the symptom. State the actual defect and why it produces the
reported behavior. A patch that makes the symptom disappear without naming the cause is not a
fix — it's a guess. Use
mgrep/ the knowledge graph to trace the failing path to its origin. - Write a minimal fix-plan. The smallest change that turns the RED test GREEN, scoped to the
cause. No refactor-while-you're-in-there — unrelated cleanup belongs to
/harness-claude:refactor-clean, not here. If the only correct fix is a design change, stop and return to/harness-claude:architector/harness-claude:planrather than hacking around it. - Add the repro as a durable regression test via
harness-claude:tdd-guide. The failing test from step 1 becomes a permanent test (RED → GREEN once the fix lands) so the bug can't silently return. This is the same TDD seam/harness-claude:implementopens each phase with. - Hand off — do not rebuild the gates. Pass the fix-plan + the RED regression test to
/harness-claude:implement(it makes the test GREEN with the minimal change), then let the change ride the existing/harness-claude:review+/harness-claude:security-review→/harness-claude:verify→/harness-claude:shipgates./harness-claude:fixdoes not duplicate review, verify, or ship.
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.
- 8d ago First seen · 93 lines · 0 tokens per session scan A bcd1b8b3be42
fix is a skill published in the GitHub repository vasuag09/harness-claude (2 stars, last pushed 2mo ago), licensed MIT. It adds 138 tokens to every session and 1,745 once invoked, about $0.0007 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
bug-fixing
Fix defects in the Composio SDK repository with focused reproduction, root-cause analysis, regression tests, and narrow verification. Use when the user reports a bug, failing test, CI regression, runtime defect, or incorrect SDK behavior. Do not use for new feature design or broad refactors.
python-testing
Select and run Python SDK verification with nox, Makefile targets, Ruff, mypy, pytest markers, sanity tests, type inference checks, and build checks. Use when adding Python tests, diagnosing Python CI, or validating Python SDK/provider changes. Do not use for TypeScript-only checks.
typescript-testing
Select and run TypeScript SDK verification for packages, examples, type checks, linting, builds, Vitest suites, and runtime E2E tests. Use when adding tests, diagnosing TypeScript CI, choosing a focused test command, or validating TypeScript package changes. Do not use for Python-only checks.
cli-e2e
Write, modify, or debug Docker-based Composio CLI end-to-end tests under ts/e2e-tests/cli, including binary invocation, fixture isolation, output assertions, and package manifests. Use for CLI E2E test suites only; use cli-command for CLI source implementation.
potpie-debug-memory
Use while debugging or troubleshooting failures, flaky tests, incidents, production alerts, CI failures, local dev setup issues, repeated bugs, prior fixes, failed attempts, and verification history.
check-cache-bugs
Audit Claude Code setup for cache bugs (CC#40524): sentinel, --resume/--continue, attribution header + ArkNill B3/B4/B5.