health
Runs a budget-aware audit of the Claude Code config stack when Claude ignores instructions, behaves inconsistently, hooks malfunction, MCP servers need auditing, or users ask why /health used many tokens. Flags issues by severity. Not for debugging code or reviewing PRs.
Security Assessment
Detected risks:
About health
The health skill is designed to conduct a budget-aware audit of the Claude Code configuration stack, particularly in scenarios where Claude may ignore instructions, exhibit inconsistent behavior, or when there are issues with hooks or MCP servers. This skill addresses the need for a structured approach to identify and flag issues based on their severity, ensuring that users can maintain the integrity and functionality of their Claude Code setup. It is not intended for debugging code or reviewing pull requests, but rather for auditing the configuration to ensure it aligns with project requirements and expectations.
Key features of the health skill include the ability to assess the project's Claude Code setup against a six-layer framework, which encompasses CLAUDE.md, rules, skills, hooks, subagents, and verifiers. The skill operates in a tiered manner, allowing it to adapt its analysis based on the complexity of the project. Users can initiate a summary audit, with the option to escalate to a more in-depth analysis if critical ambiguities arise or if the project complexity demands it. The skill also incorporates a live check of MCP servers to ensure they are operational, further enhancing the reliability of the audit process.
This skill is particularly useful for developers and teams working with Claude Code who need to ensure their configurations are functioning as intended. It is ideal for those who encounter issues with Claude's behavior or configuration settings and require a systematic way to audit and resolve these problems. By providing a clear framework for analysis and escalation, the health skill empowers users to maintain optimal performance and compliance within their projects.
FAQ
How do I initiate an audit using the health skill?
You can start an audit by running the health skill, which will automatically assess your Claude Code setup based on the defined tiers.
What are the requirements for running the health skill?
The health skill requires certain tools like jq and python3 to be installed for full functionality. If these tools are missing, some sections may show as unavailable.
Can the health skill be used for debugging code?
No, the health skill is not intended for debugging code or reviewing pull requests; it focuses solely on auditing the configuration stack.
What happens if my project exceeds the complexity threshold?
If your project is classified as complex, the health skill will perform a full six-layer setup audit and may escalate to deeper analysis as needed.
Is there a cost associated with running deep audits?
Yes, deep audits can consume significant token quota, and the skill will inform you before escalating to this level of analysis.
Install health
Quick Setup:
- Copy the skill folder to
.claude/skills/ - Claude will automatically detect and use the skill
Repository
tw93/waza