adding-models

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • Data Exposure & Exfiltration (LOW): The skill includes curl commands targeting api.letta.com. While intended for fetching model metadata, this domain is not on the pre-approved whitelist of trusted sources.
  • Indirect Prompt Injection (LOW): The skill has a data ingestion surface where it processes JSON output from an external API (api.letta.com) via jq. This presents an indirect injection surface if the external API were compromised to return malicious content.
  • Unverifiable Dependencies & Remote Code Execution (SAFE): The skill uses bun run src/index.ts for local testing. This is standard for a development skill and executes local project code rather than unverified remote scripts.
  • Command Execution (SAFE): Includes standard usage of curl, jq, and bun. These are used appropriately for the documented workflow of querying an API and running local build/test scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:19 PM