ai-provider-openai-sdk
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill contains references to non-existent or fictional AI model identifiers such as 'gpt-5.4', 'gpt-5-mini', 'gpt-5-nano', and 'o4-mini'. These entries are deceptive regarding the current actual capabilities of the OpenAI API and may lead to functional failures. Evidence is found in SKILL.md and reference.md.
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates patterns for ingesting untrusted external data (such as article text or user queries) into LLM prompts without illustrating specific sanitization or robust boundary enforcement, creating a surface for indirect injection attacks.
- Ingestion points: 'articleText' parameter in examples/structured-output.md and 'userMessage' in examples/chat.md.
- Boundary markers: Absent in code examples; input is interpolated directly into message objects.
- Capability inventory: Performs Chat Completion and Responses API calls which can influence downstream logic.
- Sanitization: No escaping or validation of the input text before it is sent to the LLM.
Audit Metadata