resonate-human-in-the-loop-pattern-python

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides documentation and examples for using the Resonate library to implement durable workflows that wait for external human input. The code examples follow official SDK patterns and do not perform any malicious operations.
  • [PROMPT_INJECTION]: The skill describes a surface for indirect prompt injection, as workflows ingest and act upon data from external sources (e.g., webhooks or CLI resolvers). In the provided examples, data from the external resolver is used to branch logic and provide context to subsequent tasks. This is a standard architectural pattern for human-in-the-loop systems. Ingestion points: The await approval statement in SKILL.md retrieves data from an external resolver. Boundary markers: Not explicitly used in the simplified documentation examples. Capability inventory: The skill demonstrates task execution via ctx.run based on the external input. Sanitization: Not shown in the basic code snippets, consistent with educational documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 06:43 AM
Security Audit — agent-trust-hub — resonate-human-in-the-loop-pattern-python