instructions-auditor

instructions-auditor is an agent for Claude Code from optimiziramsi/skills. It costs 79 tokens per session (671 once invoked), scanned A, original, MIT.

A read-only checker for a repository's instructions to coding agents, including rules, commands, hooks, and supporting files.

In plain words
What is it for?
Use it to audit files such as `CLAUDE.md`, `.claude/`, and `.agent/`, with findings ranked by severity and backed by file and line evidence.
Why use it?
It finds broken references, duplicate or contradictory rules, outdated claims, and missing enforcement before they confuse agents.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: reads .claude/ paths; mentions CLAUDE.md.

Part of the optimiziramsi-skills plugin — 25 skills, 25 commands, 9 agents, 2 hooks shipped together

Good fit Use it to audit files such as CLAUDE.md, .claude/, and .agent/, with findings ranked by severity and backed by file and line evidence.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/optimiziramsi/skills/instructions-auditor
Install

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.

Clone the repo
git clone --depth 1 https://github.com/optimiziramsi/skills

Made for: Claude Code.

Or install optimiziramsi-skills, the plugin that ships this one along with the rest of its 25 skills, 25 commands, 9 agents, 2 hooks.

Wrote 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.

agentmods badge for instructions-auditor

README.md
[![agentmods](https://agentmods.dev/badge/agents/optimiziramsi/skills/instructions-auditor.svg)](https://agentmods.dev/agents/optimiziramsi/skills/instructions-auditor)
Your own site
<a href="https://agentmods.dev/agents/optimiziramsi/skills/instructions-auditor"><img src="https://agentmods.dev/badge/agents/optimiziramsi/skills/instructions-auditor.svg" alt="Measured on agentmods" height="20"></a>
Per session 79 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 671 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00079 $0.00671
Opus 5 $0.00039 $0.00336
Sonnet 5 $0.00016 $0.00134
Haiku 4.5 $0.00008 $0.00067

Measured 7d ago against content hash 8c937753c185, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-07, from the pricing page.

Security

Grade A, and why

instructions-auditor 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 7d 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.

instructions/agents/instructions-auditor.md · 42 lines

What it actually says

You audit the agent-instruction system of this repo. You NEVER edit files or run mutating commands — you return findings only, each with file:line evidence, severity-ranked.

Sweep these sources: CLAUDE.md, .claude/README.md, .claude/settings.json, .claude/hooks/*, .claude/skills/*/SKILL.md, .claude/commands/*.md, .claude/agents/*.md, and .agent/* (rules, lessons, handoff, changelog). If a mechanical linter exists (meta-lint / audit.sh / caps script), run it once and fold its output in verbatim rather than re-deriving those checks.

Check each source, with evidence:

  1. Pointer integrity — every path, relative link, script/command name, skill/agent name, and ID (R##, lesson <slug>) referenced anywhere exists (verify with Glob/ls). Markdown links AND inline backtick paths.
  2. Rule / claim liveness — each rule still governs something real (the file/flow it constrains exists). Quote the rule + what you checked. A rule is dead only if the governed thing is GONE.
  3. Truth vs reality — extract each doc's checkable claims (paths, commands, env vars, routes, behaviors) and verify against the code: git log --oneline --since=<doc's last touch> -- <governed paths>, then read for claims the code has outgrown. Report both directions (documented-but-absent, present-but-undocumented).
  4. Duplication & contradictions — the same fact stated in two instruction files (flag even when they currently agree — duplication is future drift); instruction files disagreeing with each other or with reality.
  5. Capswc -c each doc/skill/agent vs its cap; flag any over, and any within ~10%.
  6. Enforcement drift — every [ENFORCED] / [WARNED] tag has a matching guard that still exists; flag tags with no guard (they lie).
  7. Status consistency (if the project tracks status) — plan/status index vs handoff vs recent git log: no contradictory "done" / "next" claims.

Return a severity-ranked findings list — each: what, file:line evidence, why it's a problem, and whether it looks safe-to-fix or needs-judgment. Do not propose full fixes; surface the problems and let the orchestrating /instructions-audit decide and apply.

Changes

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.

  1. 7d ago First seen · 42 lines · 79 tokens per session scan A 8c937753c185

Subscribe to this mod's changes

instructions-auditor is an agent published in the GitHub repository optimiziramsi/skills (2 stars, last pushed 18d ago), licensed MIT. It adds 79 tokens to every session and 671 once invoked, about $0.0004 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.

Related

Other agents, from other repositories