adaptive-model-routing
Pass
Audited by Gen Agent Trust Hub on Aug 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a routing mechanism to choose between different model tiers (Luna, Terra, Sol) based on task uncertainty and risk. The instructions promote security best practices by recommending escalation to more capable models for tasks involving authentication, authorization, secrets, or financial data.
- [EXTERNAL_DOWNLOADS]: The skill contains references to official OpenAI documentation and announcement pages. These are well-known, trusted services used for informational purposes and do not involve remote code execution or automated downloads of executable content.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests external data (repository rules, task descriptions) to make routing decisions, it does not demonstrate any vulnerable interpolation patterns that would allow external content to override the agent's core instructions. The risk of indirect prompt injection affecting the routing logic is considered negligible.
- [COMMAND_EXECUTION]: There are no shell commands, script executions, or system-level operations defined in the skill or its references.
Audit Metadata