using-pythia
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is to provide the agent with a decision matrix for selecting specific Oracle-related specialized skills based on natural language input (English and Vietnamese).
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a routing layer that processes user requests to trigger other tools. While it defines how to interpret untrusted user input, it does not involve unsafe interpolation into shell commands or network operations. The identified attack surface is restricted to intent classification and tool selection logic.
- Ingestion points: User requests in the session preamble (e.g., 'build/add/fix/explain X').
- Boundary markers: None specified.
- Capability inventory: Calls other pythia-related skills and local CLI tools like 'pythia guide' and 'pythia doctor'.
- Sanitization: Not explicitly defined in this router skill.
Audit Metadata