Mibo documentation
Test what your agent says and does. Mibo evaluates the traces your Agent sends against the tests you define. Find what you need below or use the search bar above.
Getting started
Section titled “Getting started”What is Mibo?How Mibo works and key concepts.
Set up your AgentSet up your Agent and send your first trace.
Quickstarts
Section titled “Quickstarts”Each quickstart takes you from sign-up to a first trace and a first test.
n8nThe verified Mibo Testing node, on n8n Cloud or self-hosted.
Vercel AI SDKAI SDK 7 telemetry exported over OTLP.
OpenLLMetryPython or TypeScript, including LangChain.
Collector fan-outKeep your current backend and add Mibo as a second exporter.
Build your test suite
Section titled “Build your test suite”Creating test casesGenerate test cases from plain language descriptions or real traces.
Your quality dashboardRead metrics and investigate evaluation results.
Trace evaluation
Section titled “Trace evaluation”Passive testingHow Mibo evaluates the traces your Agent sends.
API keys and trace routingCreate keys and configure trace routing.
Data handlingRedaction, encryption, and deletion of the data you send.
Sending traces
Section titled “Sending traces”Sending tracesDecision matrix: OTLP vs. Your API.
OpenTelemetryPoint any OTel-instrumented stack at Mibo using OTLP/HTTP JSON or protobuf.
Your APIPOST the canonical {spans:[...]} JSON directly.
n8nSend n8n traces with the community node or n8n's native OTLP export.
Flowise templateImportable Custom JS Function for any Flowise chatflow.
Reference
Section titled “Reference”Assertion referenceAll assertion types, matchers, and expression functions.
Canonical trace referenceThe {spans:[...]} structure every Mibo trace lands in.