asta-skill

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is entirely instructional and does not include any executable code, scripts, or package installations. It serves as a guide for an agent to interact with an external MCP server.
  • [SAFE]: The skill references an official MCP endpoint hosted by the Allen Institute for Artificial Intelligence (allen.ai), a well-known scientific research organization. This resource is used for legitimate academic data retrieval.
  • [SAFE]: Secret management follows security best practices by instructing the host to provide the 'ASTA_API_KEY' via environment variables rather than hardcoding credentials or requesting them in plain text.
  • [SAFE]: The instructions include defensive measures against context window exhaustion, specifically advising the agent to avoid requesting large citation or reference fields that could overflow the model's memory.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 04:04 AM
Security Audit — agent-trust-hub — asta-skill