repo-engine

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The test script tests/run-validation.py utilizes subprocess.run to execute a local companion script (scripts/validate-output.py) for automated validation of the skill's output format. The execution is handled securely by passing arguments as a list (avoiding shell injection) and using the system's own Python interpreter via sys.executable to run local files with hardcoded relative paths.
  • [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it is designed to ingest and process untrusted repository data via the read_file tool to generate onboarding artifacts.
  • Ingestion points: Files within the repository being scanned (referenced in SKILL.md and references/workflow.md).
  • Boundary markers: The skill lacks explicit prompt-level delimiters for external data, though it enforces structure through the references/artifact-schema.md and validation scripts.
  • Capability inventory: The skill is limited to list_dir and read_file as defined in the allowed-tools frontmatter in SKILL.md. It possesses no write, network, or arbitrary command execution capabilities in its primary operational loop.
  • Sanitization: Output format validation is performed by scripts/validate-output.py to ensure compliance with the repository mapping schema, and tests/validation-checklist.md instructs the agent to ground all findings in actual file evidence rather than assumptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 11:26 AM