write-good-docs
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill prioritises 'Technical truth and safety' above all other instructions, requiring that the agent does not invent or silently change technical facts, warnings, or safety information.\n- [SAFE]: Instructions in the 'Technical fidelity' reference specifically prohibit using real credentials, personal data, or production hosts in documentation examples, which helps prevent accidental data exposure.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data including repository code, existing documentation, and issue trackers.\n
- Ingestion points: SKILL.md (Workflow step 1) instructs the agent to 'inspect the relevant existing docs, templates, code, tests, configuration, issues, and pull requests' to establish facts.\n
- Boundary markers: The skill instructions require the agent to 'Separate established facts from inferences, proposals, and unknowns' and preserve technical truth as the highest priority.\n
- Capability inventory: The skill provides instructions for writing and review; it does not implement its own subprocess, file-write, or network-send capabilities.\n
- Sanitization: The 'Technical fidelity' and 'Final review' references serve as sanitisation steps by requiring factual verification against authoritative sources before finalising output.\n- [SAFE]: The skill includes references to official documentation from various technology companies and cloud services. These are well-known services and the references are informative and neutral.
Audit Metadata