health
Systematic audit of Claude Code configuration, rules, skills, hooks, and collaboration patterns.
- Detects project tier (Simple/Standard/Complex) and applies tier-appropriate checks to avoid false positives
- Audits six-layer framework: CLAUDE.md, rules, skills, hooks, subagents, and verifiers for drift and misalignment
- Runs two parallel diagnostic agents: one for context/security, one for control/behavior patterns from conversation history
- Flags critical issues (rule violations, dangerous patterns, cache breaks), structural gaps (missing layers, single-point-of-failure rules), and incremental improvements
- Outputs actionable findings with confidence levels; stops before making changes to request explicit confirmation
Health: Audit the Six-Layer Stack
Prefix your first line with 🥷 inline, not as its own paragraph.
Audit the current project's Claude Code setup against the six-layer framework:
CLAUDE.md → rules → skills → hooks → subagents → verifiers
Find violations. Identify the misaligned layer. Calibrate to project complexity only.
Output language: Check in order: (1) CLAUDE.md ## Communication rule (global over local); (2) user's recent language; (3) English.
Budget posture: Start with the summary audit. Escalate automatically when the user asks for a deep, full, complete, thorough, "深入", "完整", "彻底", or "继续跑完" audit, when current project instructions or remembered user preference says to run deep health checks by default, when the project is Complex, or when the summary pass exposes a critical ambiguity that cannot be resolved locally. Otherwise do not read full conversation extracts or launch inspector subagents. Tell the user before escalating because deep health audits can consume significant token quota.
Durable Context Preflight
Run this only when the user mentions memory, preview, previous decisions, or a prior conclusion; when they provide a memory path; or when the current project exposes an obvious local memory summary. Do not hard-code machine-specific memory roots or read raw transcripts.
Read durable context in this order: user-provided path, current project scope, then global preferences. List titles first, then open at most 1-2 relevant summaries. Treat cross-project entries as transferable patterns only.
More from tw93/claude-health
think
Turns rough ideas into approved, decision-complete plans with validated structure before writing code. Covers new features, architecture decisions, and value judgments about whether to build, keep, or remove something. Not for bug fixes or small edits.
28check
Reviews code diffs and release-ready changes after implementation, executes approved implementation plans, extracts project-specific constraints from repository context, auto-fixes safe issues, and drives approved release, publish, push, release-reaction, and issue/PR follow-through. Also triages issues and PRs when the user mentions them. Not for exploring ideas, debugging, or document prose review.
27hunt
Finds root cause of errors, crashes, regressions, screenshot-reported defects, unexpected behavior, and failing tests before applying any fix. Not for code review or new features.
26design
Produces distinctive, production-grade UI for any component, page, or visual interface. Handles screenshot-driven iteration when the user sends an image with a visual complaint. Not for backend logic or data pipelines.
26read
Fetches any URL or PDF as clean Markdown for reading, quoting, citation, or downstream work. Handles paywalls, JS-heavy pages, X/Twitter, and Chinese platforms via proxy cascade. Not for local text files already in the repo.
26learn
Runs a six-phase research workflow to turn unfamiliar domains or collected sources into publish-ready output. Not for quick lookups or single-file reads.
25