Crewcut Audit — Security skill for Claude Code
Audit the whole session's context budget — what was read, what it cost, what to do about it.
How to install Crewcut Audit
Installs to ~/.claude/skills/hzdf-2026-crewcut-crewcut-audit/SKILL.md
mkdir -p ~/.claude/skills/hzdf-2026-crewcut-crewcut-audit && curl -fsSL https://raw.githubusercontent.com/HZDF-2026/crewcut/HEAD/commands/crewcut-audit.md -o ~/.claude/skills/hzdf-2026-crewcut-crewcut-audit/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Crewcut Audit does
description: Audit the whole session's context budget — what was read, what it cost, what to do about it
Audit the current session's context as a budget, not a transcript. Report:
- Inventory — the distinct files, search results, and command outputs currently in context, grouped by the task that pulled them in.
- Dead weight — reads that entered early and have not been touched since: they are still billed on every turn. Name them.
- Live weight — reads still load-bearing
Alternatives in Security
- Octoboss Clean Contexts — You are the Octoboss — a cross-tentacle orchestrator 1.4k ★
- Imcodes — The IM for agents 960 ★
- ThinkWatch — Self-hosted AI API and MCP gateway for organizations: SSO and RBAC, per-user identity for MCP tool calls, PII 817 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
Claude Trim
Audit your Claude Code token usage — see which skills eat your context budget and find conflicts before they c
Start Impl
Start an IMPL pane session — read context and execute AUDIT directives
Claude Code Tooling Skills
Claude Code skills for keeping a Claude Code setup safe: review a third-party skill before installing it, lint
Ingest Convo
Ingest a whole past conversation into context — resolves which transcript, renders it clean via the committed
Agent Context Bench
Audit AGENTS.md, CLAUDE.md, Cursor and Copilot instructions; measure token cost and benchmark AI coding-agent
Claude Code Ledger
Local-first cost & tool-call audit for Claude Code. Finds wasted context: repeated reads, blind retries, huge
Related Agents
Factory Reviewer
Independent review of a factory solve session's whole diff before the MR, from a fresh context and read-only —
Cortex Auditor
Read-only vault auditor for the Cortex second-brain. Dispatched by the /cortex-audit ritual to scan the WHOLE
FinOps Analyst
Analyzes Northstar infrastructure cost — reads Terraform plan fixtures and environment configs, compares insta