skills/openai/plugins/reactome-skill/Gen Agent Trust Hub

reactome-skill

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • Flexible Network Operations: The scripts/rest_request.py script serves as a generic REST client. While the skill is intended to interact with reactome.org, the script's logic allows it to perform GET and POST requests to any URL provided in the input payload, including overriding the base URL if the path parameter contains a full URL.
  • File System Interaction: The skill includes a feature to save raw API responses to the local file system. Through the save_raw and raw_output_path parameters, the script can write data to specified paths or a default location in /tmp, which is a functional capability for handling large data payloads.
  • Indirect Prompt Injection Surface:
  • Ingestion points: The skill ingests data from external REST APIs (documented as reactome.org) which is then processed by the agent.
  • Boundary markers: The helper script implements a _compact function that truncates long strings and limits object depth, which provides a level of structural control, though it does not use explicit delimiters to isolate untrusted content from agent instructions.
  • Capability inventory: The skill possesses the capability to execute a local Python script, perform network requests, and write files to the disk.
  • Sanitization: Responses are parsed as JSON and truncated before being returned to the agent, reducing the risk of large-scale injection or context flooding.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:21 PM
Security Audit — agent-trust-hub — reactome-skill