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/vasuag09/harness-claudenpx agentmods add skills/vasuag09/harness-claude/extractWrote 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/extract)<a href="https://agentmods.dev/skills/vasuag09/harness-claude/extract"><img src="https://agentmods.dev/badge/skills/vasuag09/harness-claude/extract.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 | $0.00062 | $0.01096 |
| Opus 5 | $0.00031 | $0.00548 |
| Sonnet 5 | $0.00012 | $0.00219 |
| Haiku 4.5 | $0.00006 | $0.00110 |
Grade A, and why
extract 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 — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/extract — propose a skill, evaluate it, stage it
Goal: capture a repeatable workflow as a candidate skill a human can review and promote. This closes the loop the Stop detector opens — it never auto-creates a skill (the v0.1 staging rule). The output is a proposal, not a live skill.
Opt-in. No default-pipeline skill or hook invokes
/extract. Running it is always explicit and changes no baseline behavior. (v0.3 AC-E4; preserves the staging rule.)
When to run
- A Stop candidate landed in
.claude/skills-staging/candidates.md(the detector found a tool/skill subsequence that recurred this session), or - you just did something repeatable/non-obvious worth keeping.
Do this
- Read the evidence. Open
.claude/skills-staging/candidates.mdfor the detected sequence + recurrence count. Cross-check what actually fired in.claude/traces/<date>.jsonl(skills, subagents, MCPs). Skim the relevant transcript span to recover the purpose — the sequence names the steps, not why. - Draft the skill. Write a candidate
SKILL.mdwith valid frontmatter (name,description) and numbered steps that capture the workflow. Pick anamethat does not collide with an existingskills/*/SKILL.md.- Write it ONLY to
.claude/skills-staging/<slug>/SKILL.md. Never create or edit anything underskills/(or any live skill dir).
- Write it ONLY to
- Evaluate it (deterministic gate):
It scores five criteria and exits 0 (all PASS, none manual — not reachable until the AC-E3 benchmark lands) · 1 (a check FAILED) · 2 (no FAIL, MANUAL criteria remain).node scripts/eval/extract-rubric.js .claude/skills-staging/<slug>/SKILL.md- R1–R3 are deterministic (frontmatter present · ≥2 steps · name not a duplicate).
- R4 (genuinely reusable?) is always MANUAL — the script never auto-approves a skill.
- R5 (empirical value) is result-driven: if
/benchmarkproduced.claude/eval/benchmarks/<name>.jsonfor this draft, R5 reports PASS/FAIL from it; otherwise it degrades to MANUAL. Run/benchmark --component <draft-name>to close it.
- On FAIL (exit 1): fix the draft (add frontmatter, add steps, or rename to avoid the collision) and re-run. Do not stage a draft that FAILs.
- Adjudicate MANUAL (exit 2 — the expected happy path): judge R4/R5 yourself with
evidence — is this a workflow you'd actually reach for again, distinct from existing
skills? Record the verdict + reasoning in
.claude/skills-staging/<slug>/eval.mdalongside the draft. - Stage for approval. Stop there. Tell the user the proposal is staged at
.claude/skills-staging/<slug>/(draft + eval note) and that promoting it intoskills/is their explicit call. Do not promote it yourself.
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 · 69 lines · 62 tokens per session scan A b64b4e9f93d2
extract is a skill published in the GitHub repository vasuag09/harness-claude (2 stars, last pushed 2mo ago), licensed MIT. It adds 62 tokens to every session and 1,096 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 skills, from other repositories
pr-triage
4-phase PR backlog management with audit, deep code review, validated comments, and optional worktree setup. Use when triaging pull requests, catching up on pending code reviews, or managing a backlog of open PRs. Args: 'all' to review all, PR numbers to focus (e.g. '42 57'), 'en'/'fr' for language, no arg = audit…
audit-agents-skills
Audit Claude Code agents, skills, and commands for quality and production readiness. Use when evaluating skill quality, checking production readiness scores, or comparing agents against best-practice templates.
eval-agents
Audit Claude Code agents defined in .claude/agents/ for description specificity, model tier appropriateness, tools scoping, and system prompt quality. Detects dispatch ambiguity between agents, flags over-permissive tool grants, and checks for human-in-the-loop patterns that break programmatic orchestration. Use when…
check-cache-bugs
Audit Claude Code setup for cache bugs (CC#40524): sentinel, --resume/--continue, attribution header + ArkNill B3/B4/B5.
issue-triage
3-phase issue backlog management with audit, deep analysis, and validated triage actions. Use when triaging GitHub issues, sorting bug reports, cleaning up stale tickets, or detecting duplicate issues. Args: 'all' to analyze all, issue numbers to focus (e.g. '42 57'), 'en'/'fr' for language, no arg = audit only.
git-ai-archaeology
Analyze AI config evolution in a git repo. Use when mapping AI adoption history, finding when configs were first introduced, charting commit velocity by month, or identifying maturity phases in a project's AI tooling.