tech-debt
Pass
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONNO_CODE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute multiple shell commands to gather project metrics and security data, including 'git log', 'git shortlog', 'git blame', and dependency auditing tools such as 'npm audit', 'pip-audit', and 'cargo audit'.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from the analyzed codebase without adequate safety boundaries.\n
- Ingestion points: Processes all source files, comments (TODO/FIXME), and manifest files across the entire project directory.\n
- Boundary markers: Absent. The instructions do not define delimiters or specific 'ignore' rules to prevent the agent from executing instructions embedded within the codebase being analyzed.\n
- Capability inventory: Extensive filesystem read access, execution of system commands (git and audit tools), and the ability to append to local files.\n
- Sanitization: None. The skill processes and reproduces content from comments and code directly in its output report.\n- [DATA_EXPOSURE]: The 'Self-Evolution Telemetry' phase directs the agent to access the '~/.claude/projects/' directory, which is a hidden application configuration path in the user's home directory.\n- [PERSISTENCE]: The skill records execution metadata by appending to a 'skill-telemetry.md' file located in a hidden directory outside the immediate project workspace, establishing a persistent record of activities across sessions.\n- [NO_CODE]: The skill consists solely of instructions within a markdown file and does not include independent scripts or binary executables.
Audit Metadata