masterplan
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides detailed instructions for a Plan-Then-Execute architecture, ensuring implementation is grounded in repository reality. It does not include executable code, hardcoded credentials, or persistence mechanisms.
- [EXTERNAL_DOWNLOADS]: The research ledger in references/research.md contains links to research by Anthropic and Martin Fowler, as well as a community repository. These are provided for informational context and do not involve runtime downloading or execution of remote code.
- [INDIRECT_PROMPT_INJECTION]: The skill processes implementation briefs and repository files to generate plans, which is a surface for indirect prompt injection. 1. Ingestion points: Workflow step 'Load the goal' in SKILL.md. 2. Boundary markers: None specified for the brief content. 3. Capability inventory: Generates markdown plans for later execution by the pilot agent. 4. Sanitization: Not explicitly mentioned, but the skill requires verification of all plans against real repository evidence.
Audit Metadata