eva
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to route tasks to various sub-skills and reference files, creating a vulnerability surface for instruction manipulation.
- Ingestion points: Natural language user signals, as defined in the routing table in SKILL.md, are used to trigger file reads and transition execution to sub-modules.
- Boundary markers: The skill lacks explicit delimiters or specific 'ignore embedded instructions' directives for the initial parsing of user intent, which could allow instructions within the user input to influence the routing process.
- Capability inventory: The skill has the capability to read local files (references and other SKILL.md files) and delegate complex tasks to other skills like Create, Review, and Teaching.
- Sanitization: There is no documented logic for sanitizing or validating user input before it is evaluated by the routing logic to ensure it does not contain malicious instruction overrides.
Audit Metadata