reduce-reinvention
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands to execute bundled Python scripts (
reinvention_audit.pyandreuse_catalog.py) for repository analysis and catalog generation.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes content from external repository files, which creates a vulnerability to indirect prompt injection if those files contain malicious instructions.\n - Ingestion points: The auditing scripts
scripts/reinvention_audit.pyandscripts/reuse_catalog.pyread text files across the target repository filesystem.\n - Boundary markers: The generated audit reports use Markdown tables and headers to structure output, but do not implement specific delimiters or instructions to prevent the agent from interpreting embedded instructions within code snippets captured in the 'Reinvention signal lines' section.\n
- Capability inventory: The agent can execute local scripts, write audit reports to the filesystem, and analyze source code to provide architectural and consolidation recommendations.\n
- Sanitization: The scripts do not perform sanitization, escaping, or filtering of the code segments extracted from the repository before they are presented to the agent for analysis.
Audit Metadata