Docs

This quickstart takes you from a fresh account to your first tracked agent action. Pick the path that suits your stack: the SDK, the MCP server, or plain HTTP. All three send events to the same place.

Your API key is scoped to one organization. Treat it like a password and read it from an environment variable rather than hard coding it.

1. Create an API key

Open Settings in the dashboard, go to API keys, and create one for the organization you want to track. Copy it once and store it as VEYDRIA_API_KEY in your environment. You can revoke a key at any time without affecting the others.

2. Install the SDK

The TypeScript SDK works in any Node or edge runtime.

pnpm add @veydria/sdk

3. Track an agent action

Call track after your agent produces a result. Pass a stable agentExternalId so events line up with the right agent in your inventory. The call is non blocking and safe to run in production paths.

import { Veydria } from '@veydria/sdk';

const veydria = new Veydria({ apiKey: process.env.VEYDRIA_API_KEY });

await veydria.track({
  agentExternalId: 'support-bot',
  input: userMessage,
  output: modelReply
});

4. Connect over MCP

If your agent speaks the Model Context Protocol, add the Veydria MCP server to your client config. It reports actions in for you, so there is no custom code to maintain.

{
  "mcpServers": {
    "veydria": {
      "command": "npx",
      "args": ["-y", "@veydria/mcp"],
      "env": { "VEYDRIA_API_KEY": "your_key_here" }
    }
  }
}

5. Send events with plain HTTP

No SDK on your platform? Post to the ingest endpoint with a bearer token. The body matches the SDK payload, so you can move between the two later without changing your data.

curl -X POST https://veydria.devpilotx.com/api/ingest \
  -H "Authorization: Bearer $VEYDRIA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "agentExternalId": "support-bot",
    "input": "How do I reset my password?",
    "output": "Open Settings, then Security, then Reset password."
  }'

Next steps

Once events are flowing, classify the system to see its risk tier, run an evaluation, and generate your first document. Each step reads from the data you just connected.