fable-plan
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests 'requirements' and 'evidence' to generate architectural plans. This ingestion of untrusted external data creates a potential surface for indirect prompt injection.\n
- Ingestion points: Data enters the agent context through the
requirementsandevidenceinput fields defined inSKILL.md.\n - Boundary markers: The skill relies on architectural invariants and structural templates but lacks explicit instructions to sanitize or disregard commands embedded in input data.\n
- Capability inventory: The skill manifest (
skill.package.json) defines no executable scripts, and the skill is restricted to a 'data-only' policy.\n - Sanitization: No explicit data sanitization or filtering logic is specified for the input streams.\n- [SAFE]: No evidence of malicious behavior, such as credential theft, obfuscated network calls, or unauthorized system modifications, was detected. The skill is configuration-oriented and adheres to modularity best practices.
Audit Metadata