skill-stocktake
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local bash scripts (scan.sh, quick-diff.sh, save-results.sh) to manage the inventory and results. These scripts are part of the skill package and use jq to handle data safely.
- [PROMPT_INJECTION]: The skill audits third-party skill files, creating an indirect prompt injection surface. Ingestion points: scripts/scan.sh reads files in .claude/skills/. Boundary markers: The subagent prompt uses [INVENTORY] and [CHECKLIST] delimiters. Capability inventory: Subprocess calls for bash scripts and subagent execution. Sanitization: Results are processed and formatted using jq to prevent injection into JSON reports.
- [SAFE]: The skill accesses ~/.claude/observations.jsonl to provide usage statistics. This is a local configuration file and the data is used solely for frequency counting without external transmission.
Audit Metadata