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

bindingdb-skill

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [Generic REST Client Capability]: The scripts/rest_request.py script acts as a general-purpose HTTP client. While the skill is intended for BindingDB interactions, the implementation allows for GET and POST requests to any base_url. This capability could potentially be used for interactions with external or internal network resources if the agent is directed to unintended destinations.\n- [Local File Writing]: The skill includes functionality to save raw API responses to the local file system using the save_raw and raw_output_path parameters. The script is designed to create necessary directories and write the response content to the provided path, which allows the skill to persist data to the environment.\n- [Indirect Prompt Injection Surface]: The skill retrieves and processes content from external APIs. As with any skill that ingests external data, there is a possibility that untrusted content could include instructions that influence the agent's context and subsequent behavior.\n
  • Ingestion points: External data enters the agent context via network requests performed in scripts/rest_request.py.\n
  • Boundary markers: The skill does not currently use specific delimiters or instructions to isolate external data from the agent's primary directives.\n
  • Capability inventory: The skill is capable of performing network requests and writing data to the local file system.\n
  • Sanitization: The skill processes JSON or text responses without specific filtering or sanitization of potential instruction-like content.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 12:20 PM
Security Audit — agent-trust-hub — bindingdb-skill