openai-api
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill manages dependencies via standard package managers and interacts with official OpenAI API endpoints.
- Fetches the
openaiSDK and development tools from the official npm registry intemplates/package.json. - Communicates with official OpenAI API hosts including
api.openai.comandwss://api.openai.comacross multiple templates (e.g.,templates/chat-completion-basic.ts,templates/streaming-fetch.ts,templates/rate-limit-handling.ts). - [COMMAND_EXECUTION]: Includes a utility script for local environment management.
- The script
scripts/check-versions.shexecutesnpm listandnpm viewto compare installed package versions with those available on the registry. - [INDIRECT_PROMPT_INJECTION]: The skill provides templates for processing external data, which introduces a common surface for indirect prompt injection.
- Ingestion points: User-supplied strings, images, and audio files are ingested in files like
templates/chat-completion-basic.ts,templates/vision-gpt4o.ts, andtemplates/audio-transcription.ts. - Boundary markers: The templates do not utilize explicit boundary markers or system instructions to ignore embedded commands in the processed data.
- Capability inventory: The skill possesses network capabilities to send data to OpenAI and file-writing capabilities to save audio/image outputs (e.g.,
templates/text-to-speech.ts). - Sanitization: The provided code examples do not include explicit sanitization or filtering logic for the content sent to the model endpoints.
Audit Metadata