sldd
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions in
steps/feature/05-implementation-green.mddirect the agent to follow "repository instructions" and "agentic instructions present in context" during implementation. Ingestion points: Data read from the project repository or provided in the agent's context (e.g., project-local instructions). Boundary markers: The skill does not define specific delimiters or "ignore" instructions for these external inputs to prevent malicious command execution. Capability inventory: The skill grants the agent the ability to modify production code and execute test commands in Step 04 and Step 05. Sanitization: No explicit sanitization, filtering, or validation of the content of these context-provided instructions is implemented.
Audit Metadata