Quick Start

Get started in 2 minutes.

Step 1: Get Your API Key

1. Sign up at converra.ai
2. Go to Integrations → API Keys
3. Create and copy your key

Step 2: Choose Your Integration

Option A: AI Assistant (MCP)

Open Cursor, Claude Code, Windsurf, or any AI coding assistant and say:

"Install the Converra MCP server with API key [paste your key here], then help me upload my agents and optimize them."

That's it. Your AI handles the setup.

Option B: SDK (Programmatic)

Wrap your LLM client with one line — all conversations are captured automatically.

Node.js

npm install converra

Python

pip install converra

Node.js

import { Converra } from 'converra';
import OpenAI from 'openai';

const converra = new Converra({ apiKey: 'sk_live_...' });
const openai = converra.wrap(new OpenAI());

// Use openai normally — conversations captured automatically
const response = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello' }],
});

Python

from converra import Converra
from openai import OpenAI

converra = Converra(api_key="sk_live_...").init()
client = converra.wrap(OpenAI())

# Use client normally — conversations captured automatically
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
)

See the SDK Guide for Anthropic, Vercel AI SDK, and auto-instrumentation.

Option C: Zero-Code Auto-Instrumentation

No code changes needed — just set environment variables:

CONVERRA_API_KEY=sk_live_... node --import converra/auto server.js

Every OpenAI and Anthropic call is captured automatically.

What You Can Do

Once connected, Converra:

• Diagnoses why your agents underperform
• Generates targeted fixes (prompt, config, model)
• Tests every fix against 36+ simulated conversations
• Ships proven improvements automatically

With MCP, just ask:

• "Analyze my support agent for issues"
• "Optimize my onboarding agent"
• "Compare my current agent vs this new version"

---

Manual MCP Setup (Optional)

Claude Code

claude mcp add converra -- npx -y mcp-remote@latest https://converra.ai/api/mcp \
  --header "Authorization:Bearer YOUR_API_KEY"

Cursor

Add to .cursor/mcp.json:

{
  "mcpServers": {
    "converra": {
      "url": "https://converra.ai/api/mcp",
      "headers": { "Authorization": "Bearer YOUR_API_KEY" }
    }
  }
}

Other Platforms

See MCP Setup Guide for Claude Desktop, Windsurf, and AntiGravity.

---

Import from LangSmith or Langfuse

Already using an observability platform? Connect for continuous sync of traces and feedback.

1. Go to Integrations in the Converra app
2. Click Connect LangSmith (or Connect Langfuse) and paste your API key
3. Select your workspace and project
4. Configure sync frequency (hourly to daily)
5. Converra discovers your agents and agent systems automatically

Multi-agent traces are grouped into agent systems with path visualization and weakest-link scoring.

See the LangSmith or Langfuse integration guides.

---

Next Steps

• SDK Guide - LLM client wrapping, tracing, A/B testing
• Integrations Overview - All connection options
• MCP Tools Reference - All MCP commands
• REST API - Direct API integration