generate-markdown
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes external narrative content from Sanity CMS (Portable Text and modules), which creates a surface for indirect prompt injection where malicious instructions in the source content could attempt to influence agent behavior.
- Ingestion points: Content is ingested from the Sanity CMS using
get_documentandquery_documentstools (SKILL.md, Step 1). - Boundary markers: The instructions include a strict verbatim constraint: 'Transcribe verbatim — no paraphrasing, no reordering, no dropped sentences' (SKILL.md, Step 2).
- Capability inventory: The skill can read local codebase files (
src/modules/*/schema.ts,src/lib/env.ts) and modify CMS data viapatch_documents(SKILL.md, Step 6). - Sanitization: The skill relies on the verbatim transcription directive to prevent the execution of embedded instructions rather than using automated sanitization.
Audit Metadata