api-writer
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an inherent attack surface for indirect prompt injection due to its workflow of reading and interpreting external requirement documents to generate code-like contracts.
- Ingestion points: As defined in Phase 0 of
SKILL.md, the agent uses theGlobtool to scan and collect context from PRDs, requirement documents, and existing API specifications. - Boundary markers: The instructions implement a safety barrier by requiring the agent to use
AskUserQuestionto have the user confirm the document baseline and document paths before any content is read or acted upon. - Capability inventory: The skill's primary capabilities involve reading file contents and outputting structured markdown documentation based on predefined templates. It does not contain instructions to execute arbitrary code or perform network exfiltration.
- Sanitization: The skill relies on strict template-based output and the principle of 'based on evidence, not guessing' to reduce hallucinations or unintended instruction following, though it lacks explicit input sanitization logic.
Audit Metadata