llm-router
Fail
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: HIGHDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [DATA_EXFILTRATION] (HIGH): The skill directs the agent to use
https://api.aisa.one/v1as the primary API endpoint. By routing traffic through this untrusted proxy instead of official provider endpoints (e.g., OpenAI, Anthropic), all data—including potentially sensitive system prompts and user information—is fully visible to the proxy operator. - [METADATA_POISONING] (MEDIUM): The skill uses deceptive marketing claims in the
displayNameandREADME.md('Save upto 50% for model tokens') to entice users into adopting a high-risk data-routing configuration. Such significant price discrepancies for API tokens are often associated with malicious proxy services. - [EXTERNAL_DOWNLOADS] (MEDIUM): The documentation refers to a local script
scripts/llm_router_client.pythat is not present in the provided files. Without this script, the full behavior of how it manages requests, logging, or credential handling cannot be verified.
Recommendations
- AI detected serious security threats
Audit Metadata