cohere-core-workflow-b

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The implementation of the agent loop in SKILL.md creates a surface where malicious data from external tools could influence agent behavior.\n
  • Ingestion points: Data returned by functions in toolExecutors (such as search_database) is ingested into the conversational history in the agentLoop function.\n
  • Boundary markers: The implementation lacks delimiters or system instructions that would help the model distinguish between legitimate tool outputs and malicious instructions embedded within those outputs.\n
  • Capability inventory: The skill's environment configuration permits the use of Write and Bash tools, which provides a high-impact execution path if the agent is successfully injected.\n
  • Sanitization: The agentLoop directly pushes raw tool results into the messages array via messages.push({ role: 'tool', toolCallId: tc.id, content: result }) without any validation or escaping of the content string.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 04:15 PM
Security Audit — agent-trust-hub — cohere-core-workflow-b