plan-milestone
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from various markdown files which could potentially contain hidden instructions intended to influence the agent's behavior.
- Ingestion points: Processes content from
docs/specifications/product.md,AGENTS.md,CONTEXT.md, and architectural decision records (ADRs) as specified inSKILL.md. - Boundary markers: Absent; the skill does not instruct the agent to use specific delimiters or to disregard instructions embedded within these files.
- Capability inventory: Capabilities are restricted to file system read/write operations on markdown files in the
docs/directory. There is no network access or command execution capability. - Sanitization: No explicit sanitization or validation of the ingested documentation content is performed.
Audit Metadata