detail
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script to analyze the repository structure.
- Evidence: The process explicitly calls
python3 scripts/codebase-map.py . --format markdownto generate codebase maps during the planning phase. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the user's repository which could contain malicious instructions.
- Ingestion points: Reads
docs/arc/specs/*-spec.md, vetted findings, package manager lockfiles, and direct user requests inSKILL.md. - Boundary markers: The skill uses normative XML tags for task granularity and specifies a structured plan schema, which provides some structural separation between instructions and data.
- Capability inventory: The skill has the capability to execute local scripts (
scripts/codebase-map.py), write files todocs/arc/plans/, and modify the plan index. - Sanitization: The process includes a mandatory validation step using a
plan-document-reviewer.mdagent to verify the plan against canonical contract dimensions.
Audit Metadata