model-id-lookup
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill fetches model definitions from
https://models.dev/model-schema.json. This domain is not among the pre-approved trusted sources, and the downloaded data is used to overwrite a local file (references/model-schema.json). - COMMAND_EXECUTION (LOW): The skill instructions direct the agent to use the
grepcommand for searching. While the intent is a local search, if user-provided keywords are concatenated directly into the shell command without sanitization, an attacker could achieve command injection (e.g., searching for\"; rm -rf /; #\"). - INDIRECT_PROMPT_INJECTION (LOW): The skill processes untrusted remote data which is then displayed to the agent.
- Ingestion points:
update_models.py(via urllib) andSKILL.md(via webfetch). - Boundary markers: None identified; external content is parsed directly as JSON.
- Capability inventory:
grep(shell execution),write(file system write),webfetch(network access). - Sanitization: No validation or sanitization of the string content within the JSON enum is performed before searching.
Audit Metadata