fable-research
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [SAFE]: The skill operates as a set of logical instructions for evidence gathering and does not include any executable scripts or high-risk tool permissions. It explicitly instructs the agent to favor primary, authoritative sources over unverified secondary sources, reducing the risk of acting on malicious or incorrect information.\n- [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by processing external primary sources and library documentation.\n
- Ingestion points: Untrusted data enters the agent context through the processing of external primary sources and research queries as defined in SKILL.md.\n
- Boundary markers: Stage 5 of the research protocol requires a clear separation between 'Fact' and 'Interpretation,' which serves as a logical boundary for processing external data.\n
- Capability inventory: The skill is configured as 'data-only' and possesses no local scripts, command execution capabilities, or sensitive file system access tools within its own configuration.\n
- Sanitization: No technical sanitization of external text is mentioned, but the protocol prioritizes normative specifications and official source code to minimize exposure to lower-tier anecdotal content. The risk is considered safe given the skill's primary research purpose and lack of dangerous internal capabilities.
Audit Metadata