models-dev-query
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches aggregated AI model metadata, including pricing and context limits, from
https://models.dev/api.json. This is a well-known community resource. - [EXTERNAL_DOWNLOADS]: Retrieves detailed model specifications (TOML files) from the official
anomalyco/models.devrepository on GitHub using the GitHub CLI (gh api) andcurl. GitHub is a trusted service. - [COMMAND_EXECUTION]: Uses local shell commands including
curl,jq,gh,sed,xargs, andsortto query, filter, and format model information for presentation to the user. - [PROMPT_INJECTION]:
- Ingestion points: Ingests external model specification data from
models.devand GitHub (SKILL.md). - Boundary markers: None explicitly present in the provided instructions; processed content is passed directly to the agent context.
- Capability inventory: Data retrieval and formatting via
curl,gh, andjq. No file system writes or code execution capabilities are present. - Sanitization: External data is processed via structured query tools (
jq), which mitigates direct code injection but maintains a surface for indirect prompt injection if the source data contains malicious natural language instructions.
Audit Metadata