The Agent Skills Directory

[SAFE]: No security issues detected. The skill performs local file system reads and basic command execution to assess project health.
[DATA_EXPOSURE]: The skill accesses sensitive local files including ~/.claude/usage.jsonl and connections.md. This access is necessary for the skill's primary purpose of auditing usage patterns and infrastructure configuration. No evidence of network exfiltration was found; all output is directed to local documentation files.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from multiple ingestion points which creates a potential attack surface.
Ingestion points: Files within context/, connections.md, skills/, commands/, agents/, recipes/, decisions/log.md, and ~/.claude/usage.jsonl (SKILL.md).
Boundary markers: Absent; there are no specific delimiters or instructions to ignore embedded commands within the audited files.
Capability inventory: File reading (ls, git status), tool inspection (claude mcp list), and local file writing to docs/audits/ and decisions/log.md (SKILL.md).
Sanitization: Absent; the skill does not explicitly filter or escape content read from files before incorporating it into the audit report.

audit