ask-howells
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to read and verify information against local SKILL.md files from other installed skills while processing user situational descriptions. This creates a potential surface for indirect prompt injection where instructions embedded in those external files could attempt to manipulate the agent's responses.\n
- Ingestion points: Processes user situational descriptions and local SKILL.md files from the filesystem (referenced in Rules 2 and 3).\n
- Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands within the ingested skill documentation.\n
- Capability inventory: The skill is restricted by
disable-model-invocation: trueand explicitly instructs the agent not to start work, open project files, or execute other skills (Rule 1). The agent's actions are limited to providing text-based recommendations.\n - Sanitization: No evidence of sanitization, filtering, or validation of the content from the external skill files was found.
Audit Metadata