blaxel

Warn

Audited by Snyk on Feb 26, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

  • Potentially malicious external URL detected (high risk: 0.80). The skill instructs agents to call a Blaxel-hosted MCP endpoint at runtime (e.g., https://run.blaxel.ai/{workspace}/functions/{name}/mcp and via bl_tools like await bl_tools(['sandbox/my-sandbox'])), which fetches tool definitions at runtime that directly affect agent behavior/instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 26, 2026, 02:44 PM