analyze-repo
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands for repository inspection, metrics collection, and to open the generated report using platform-native utilities like 'open', 'xdg-open', or 'start'.\n- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted repository data, which could contain malicious instructions designed to influence agent behavior. This risk is addressed through mandatory sanitization requirements.\n
- Ingestion points: Analysis of repository source files, configuration, and command outputs as described in Step 1 of the workflow.\n
- Boundary markers: Present in the 'Encoding contract' section of 'references/html-report-template.md', which warns to treat repository content as untrusted.\n
- Capability inventory: Shell command execution (metrics, platform openers), extensive file system read access, and local file writing for report generation.\n
- Sanitization: The skill explicitly mandates HTML and XML escaping for all untrusted repository text and labels before inclusion in the output.
Audit Metadata