cohere-api

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a vulnerability to indirect prompt injection in its RAG implementation pattern.\n
  • Ingestion points: Untrusted data enters the context through the docs array in the RAG implementation example in SKILL.md.\n
  • Boundary markers: No delimiters or instructions to ignore embedded commands are present; the data is joined directly into the prompt string.\n
  • Capability inventory: The skill uses the client.generate method to call Cohere's generation endpoint, which can produce text based on the injected instructions.\n
  • Sanitization: No escaping or validation of the retrieved document content is performed before interpolation.\n- [EXTERNAL_DOWNLOADS]: The documentation instructs the user to install the cohere package via pip. This is the official SDK for the Cohere service, a well-known AI provider.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 10:46 PM