openrouter-trending-models
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests model metadata and ranking information from an external provider (OpenRouter) which is then used by the agent for model selection and documentation updates.
- Ingestion points: The script
scripts/get-trending-models.tsfetches data fromhttps://openrouter.ai/rankingsandhttps://openrouter.ai/api/v1/models. - Boundary markers: The skill output is structured JSON, providing a schema boundary, but lacks explicit instructions for the agent to disregard potential instructions embedded in external strings like model names or IDs.
- Capability inventory: The skill can execute shell commands via
bun run, perform network requests, and write to the local file system. - Sanitization: While the script parses external data into a defined JSON structure, it does not explicitly sanitize the text content of the fields before they are presented to the agent.
- [EXTERNAL_DOWNLOADS]: Fetches current AI model rankings, usage statistics, and pricing details from OpenRouter's official website and API.
- [COMMAND_EXECUTION]: The skill relies on the execution of a local Bun script (
scripts/get-trending-models.ts) to retrieve and process ranking data.
Audit Metadata