openrouter-model-availability

Warn

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: MEDIUMPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PERSISTENCE]: The skill provides an 'Availability Monitoring Script' and instructs the user to configure it as a cron job (*/5 * * * * /path/to/check_models.sh) to automate model availability checks. This establishes a mechanism for recurring script execution on the host to maintain monitoring activity.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the external OpenRouter API (including the model catalog and chat completion results) which is then processed and displayed by the agent.\n
  • Ingestion points: Data retrieved from the /api/v1/models endpoint and responses from model health probes in SKILL.md and references/status-dashboard.md.\n
  • Boundary markers: No delimiters or explicit instructions are provided to the agent to treat the API-returned data as untrusted or to ignore instructions embedded within the strings.\n
  • Capability inventory: The skill is granted access to the Bash tool (configured for python3, curl, and jq), as well as Read, Write, and Edit permissions.\n
  • Sanitization: The skill relies on basic JSON parsing but lacks robust sanitization for natural language fields within the API responses that could be manipulated to influence the agent's behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 9, 2026, 03:46 AM
Security Audit — agent-trust-hub — openrouter-model-availability