codebase-reverse-spec

Pass

Audited by Gen Agent Trust Hub on Jul 8, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data by reading implementation details from existing codebases during Phase 2. This creates a surface for indirect prompt injection where malicious instructions embedded in code comments could influence the agent's behavior.
  • Ingestion points: Phase 2 (Rule extraction) reads files from the user-provided codebase.
  • Boundary markers: Absent. The skill does not define specific delimiters or instructions to ignore embedded directives in the analyzed files.
  • Capability inventory: The skill can list directories (Glob), read files, and execute local shell scripts.
  • Sanitization: Absent. The content read from files is not filtered or sanitized before being processed.
  • [COMMAND_EXECUTION]: The skill includes and instructs the execution of a local shell script, scripts/scan_leakage.sh, during Phase 4 to identify technical jargon in the output. While the script's content is focused on benign pattern matching using grep, the practice of executing local scripts provided with a skill should be monitored for potential misuse or argument injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 8, 2026, 10:10 PM
Security Audit — agent-trust-hub — codebase-reverse-spec