frontend-skills-stats
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes several bash commands, including git, jq, ls, and wc, to gather metadata about installed hooks and aggregate performance metrics. It also calls a project-local script scripts/generate-hook-configs.sh with the --check flag to identify configuration drift between the manifest and the environment.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill reads operational data from local files, specifically session metrics in ~/.claude/hook-metrics/ and configuration files like skill-manifest.json and .claude/settings.json. This data is used solely for generating the requested report and is not transmitted externally or used to access sensitive user credentials.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from JSON metric files, which serves as a potential ingestion surface for indirect prompt injection.
- Ingestion points: Files in ~/.claude/hook-metrics/ are parsed via jq and analyzed by the agent.
- Boundary markers: The instructions do not specify boundary markers to isolate the parsed data from the agent's command context.
- Capability inventory: The skill has access to local shell execution and file system reads within the repository and the ~/.claude directory.
- Sanitization: The content is processed using standard JSON parsing without explicit sanitization steps; however, this behavior is consistent with the primary purpose of an analytics tool.
Audit Metadata