tome
Pass
Audited by Gen Agent Trust Hub on Oct 28, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze potentially untrusted data sources to produce technical documentation and articles.
- Ingestion points: Processes git diffs, PR descriptions, commit histories, interview transcripts, podcasts, and user-provided rough drafts (described in
SKILL.mdandreference/article-handoffs.md). - Boundary markers: The skill requires the use of
[Inference: evidence]tags andLOW CONFIDENCEflags to distinguish interpreted content from technical facts. It also includes instructions to mask internal-only details before publication. - Capability inventory: The agent has the ability to create and modify markdown files, update project indices, and generate content for external platforms.
- Sanitization: The instructions explicitly mandate protecting internal context, masking credentials, and excluding security implementation details from learning materials.
Audit Metadata