Skip to main content
Runnable examples show how to give your agent a simple capability: when it notices friction, confusion, bugs, feature requests, or delight, it can post a structured observation to Correl8. Repository: github.com/Correl8AI/examples

​
Prerequisites

  1. A Correl8 workspace and project (app.correl8.ai — free plan is enough)
  2. A project API key — Settings → create/copy a key → CORREL8_API_KEY
  3. CORREL8_PROJECT_ID from your project URL: /project/{project_id}/...
  4. An OPENAI_API_KEY
Each example includes its own .env.example. Copy it to .env before running. The same values work across examples.

​
Quick starts (CLI)

ExampleIntegrationStart here
basic-agentREST custom toolYes
langchain-agent-observationsLangChain + MCP
openai-agents-sdkOpenAI Agents SDK + MCP
pydantic-ai-agentPydantic AI + MCP

​
Full demo app

ExampleDescription
movie-recommendationBrowser chat app → agent → post_observations → Correl8

​
Verify

After running an example, check Observations in the app: 
https://app.correl8.ai/project/{CORREL8_PROJECT_ID}/observations
See Quickstart for the MCP setup path, or Connect overview for integration options.