openai-docs-skill

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill fetches documentation content from a remote server (developers.openai.com) and incorporates it into the agent's context. While the source is a trusted service, the ingestion of external data represents a standard attack surface for indirect prompt injection.
  • Ingestion points: Data is retrieved via curl in scripts/openai-docs-mcp.sh and returned to the agent context through subcommands like fetch and search.
  • Boundary markers: No specific delimiters or instructions to ignore embedded prompts are present in the skill's instructions.
  • Capability inventory: The skill uses curl and jq for network communication and JSON parsing.
  • Sanitization: The script extracts and returns the raw text content from the server response without additional filtering.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 04:33 AM
Security Audit — agent-trust-hub — openai-docs-skill