fable-opus
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a pattern where untrusted user input is passed to a tool-enabled sub-agent. 1. Ingestion points: User tasks and session context are passed to the @fable-orchestrator agent (SKILL.md). 2. Boundary markers: Not explicitly defined in the routing instructions; the agent is briefed with the task directly. 3. Capability inventory: The target sub-agent (@fable-orchestrator) utilizes a Bash shell for verification commands (SKILL.md). 4. Sanitization: The skill relies on sub-agent instructions forbidding file writes via Bash rather than explicit input sanitization.
- [SAFE]: No malicious patterns such as remote code execution, credential theft, or obfuscation were found. The skill serves as a structural guide for agent delegation within a managed model routing system.
Audit Metadata