Workflow-first
n8n AI agent testing
Run active tests against an n8n workflow and send execution traces through the Testing Node, HTTP Request node, or built-in OTLP tracing on supported self-hosted versions.
View the n8n workflow →Connect the agent stack you already run.
Mibo keeps active testing and passive trace evaluation together while meeting your system where it is. Start with a maintained n8n or Flowise setup, or use a framework-neutral OpenTelemetry pipeline.
Choose the path that matches your stack
The integration decides how Mibo receives or drives your agent. The test definitions and passive evaluation model stay consistent across these paths.
Framework-neutral
OpenTelemetry AI agent testing
Point an OTLP/HTTP JSON exporter at Mibo and evaluate traces from an OTel-instrumented agent, app server, LLM SDK, or workflow tool.
View the OpenTelemetry path →Managed or self-hosted
Flowise agent tracing
Send traces from Flowise Cloud or self-hosted with the maintained Mibo Trace Sender template.
View the Flowise guide →Need a different transport?
Mibo also accepts its canonical {spans:[...]} trace shape over HTTP for custom agents and managed platforms. Use the documentation for the exact payload and routing rules.
Keep exploring
AI agent evaluation
Understand the semantic and procedural checks behind every result.
Explore evaluation →Production testing
Use the same tests to evaluate real interactions in production.
Explore production testing →Mibo documentation
Follow setup instructions and reference payloads for each ingestion path.
Open the docs →