alchemy-openapi-skill

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches a curated OpenAPI JSON schema from a public GitHub repository (raw.githubusercontent.com) to define the structure of its API operations.
  • [COMMAND_EXECUTION]: The skill uses the uxc CLI to manage credential bindings and perform network requests, which involves executing shell commands in the agent's environment.
  • [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes data retrieved from an external service. 1. Ingestion points: The skill reads JSON data from api.g.alchemy.com. 2. Boundary markers: There are no explicit instructions to the agent to ignore or delimit instructions potentially embedded within the pricing data. 3. Capability inventory: The skill's capabilities are restricted to network-based read operations via the uxc tool. 4. Sanitization: The skill does not define specific validation or sanitization logic for the API responses before they are presented to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 03:36 AM