documenter
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository files like code, schemas, and tests to generate documentation, creating a potential surface for indirect prompt injection. 1. Ingestion points: Core Workflow Step 2 in SKILL.md specifies reading code, config, schemas, tests, and command output. 2. Boundary markers: SKILL.md contains instructions to reconcile conflicts, label disputed claims, and pause on material choice. 3. Capability inventory: SKILL.md authorizes execution of builds, formatters, and linters. 4. Sanitization: SKILL.md requires inspection of validation commands and verification of authority before execution.
- [COMMAND_EXECUTION]: The skill allows for the execution of documentation-related tools such as formatters and link checks. SKILL.md Step 5 (Validate proportionally) mitigates risk by requiring the agent to inspect commands for side effects and ensure proper authorization before execution.
Audit Metadata