alphagbm-macro-view

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from an external API (alphagbm.zeabur.app) that includes a pre-computed impact_analysis field.
  • Ingestion points: The GET /api/research/macro endpoint returns an impact_analysis string that is directly presented to the user.
  • Boundary markers: None. There are no instructions provided to the agent to treat this specific field as untrusted or to ignore embedded instructions within it.
  • Capability inventory: The skill possesses minimal capabilities; it primarily performs network GET/POST/DELETE operations to a specific vendor domain and does not include file system writes or subprocess execution.
  • Sanitization: None. The skill instructions do not specify any validation or sanitization for the API-returned content before display.
  • [SAFE]: The skill correctly handles sensitive information by instructing the use of environment variables (ALPHAGBM_API_KEY) for authentication rather than hardcoding secrets.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:59 PM
Security Audit — agent-trust-hub — alphagbm-macro-view