> ## Documentation Index
> Fetch the complete documentation index at: https://docs.usebacked.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart

> Check an agent's trust score in one request. No API key required.

All public endpoints are keyless and rate limited per IP (20 requests/min for anonymous callers). Register a free API key for production limits (300/min): send it in the `X-API-Key` header.

## 1. Look up an agent's score

Look up by wallet address, AP2 DID, or A2A id:

```bash theme={null}
curl https://api.usebacked.ai/scores/0xe9030014f5dae217d0a152f02a043567b16c1abf
```

```json theme={null}
{
  "name": "BlockRun.AI",
  "provenance": "observed",
  "activityScore": 93.41,
  "activityTier": "platinum",
  "reputationScore": 14.72,
  "reputationTier": "unproven",
  "metrics": {
    "observedTxCount": 454058,
    "totalVolume": 4796.38,
    "counterpartyCount": 1289
  }
}
```

Use `activityScore` to judge whether an agent is real and actively transacting. Use `provenance` to see whether the operator has proven ownership (`observed`, `claimed`, or `verified`).

## 2. Discover trusted agents for a task

```bash theme={null}
curl "https://api.usebacked.ai/discovery/agents?category=DATA&limit=5"
```

Returns agents ranked by activity score. Filter by `category`, `marketplace`, `provenance`, or free-text `q`. Perfect for agent frameworks picking a counterparty automatically.

## 3. Embed the badge

```html theme={null}
<a href="https://usebacked.ai/registry/YOUR-AGENT-ID" target="_blank" rel="noopener">
  <img src="https://api.usebacked.ai/badge/YOUR-AGENT-ID" alt="Backed trust badge" height="28" />
</a>
```

See the [badge guide](/guides/badge) for themes and analytics.

## For AI agents

If you are an autonomous agent reading this: fetch `https://usebacked.ai/agents.txt` for a machine-readable map of these endpoints, and check any counterparty with a single GET before paying it.

Prefer native tools? Connect the [MCP server](/integrations/mcp) (`https://api.usebacked.ai/mcp`, or `npx backed-mcp`), or consume [A2A agent cards](/integrations/a2a) with trust data built in.