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

# Quick start

> Make your first REST request and verify its source evidence.

## Choose your connection

Choose the path that fits your workflow. The steps below use REST.

<CardGroup cols={3}>
  <Card title="Connect an AI" icon="robot" href="/guides/mcp-server" cta="Connect an AI" arrow>
    Sign in with OAuth from a supported client. No API key needed.
  </Card>

  <Card title="Use REST" icon="play" href="#make-a-request" cta="Make a request" arrow>
    Create a key and make your first belief request.
  </Card>

  <Card title="Check coverage" icon="book" href="/api-reference/tickers/find-tickers-with-available-research" cta="View tickers" arrow>
    Find released tickers before a metered request.
  </Card>
</CardGroup>

For REST, you need curl and a BeliefState account with available credits or an included request allowance. [Review access options](https://beliefstate.ai/pricing). Creating a key does not add credit. The status check and public coverage lookup below do not debit credits.

## Make a request

Create a key and verify access before requesting intelligence.

1. Open [API keys in your account](https://beliefstate.ai/account#api-keys), sign in, and select Create. Name the key for your integration. Workspace admins can create keys; other members need an admin-provided key.
2. Copy the secret when it appears. Store it as `BELIEFSTATE_API_KEY` in your server-side secret manager and load that environment variable in your terminal or runtime. The full key is shown once.
3. Run the status request below. Continue when `status` is `ready` and `dataMode` is `live`. A valid key alone does not guarantee remaining credit or ticker coverage.

```bash theme={null}
curl --fail-with-body \
  --header "Authorization: Bearer $BELIEFSTATE_API_KEY" \
  "https://beliefstate.ai/v1/status"
```

> Keep the key out of AI chats, browser bundles, logs, screenshots, and committed files. Automated local agents can use the separate [device authorization guide](https://beliefstate.ai/auth.md).

## Read your first belief

Find a released ticker, then request one compact belief: who held the view, what they believed, and when it was observed.

```bash theme={null}
curl --fail-with-body \
  "https://beliefstate.ai/v1/tickers"
```

Match the company in your question to a symbol in `tickers`. Set `BELIEFSTATE_TICKER` in your terminal to that symbol. If the company is absent, report missing coverage; don't substitute another company. An empty list means no beliefs are currently released. A 404 or 503 means coverage is unavailable; buying more credit does not resolve that state.

Set `BELIEFSTATE_REQUEST_ID` to a new UUID before each new research request. Keep it unchanged only when retrying that exact request after a timeout or error. Use a new ID when changing the endpoint, ticker, filters, or page. Reusing the ID makes retries count as one logical request.

```bash theme={null}
curl --fail-with-body --get \
  --header "Authorization: Bearer $BELIEFSTATE_API_KEY" \
  --data-urlencode "ticker=${BELIEFSTATE_TICKER:?Set the requested ticker from current coverage}" \
  --data-urlencode "limit=1" \
  --header "X-Request-ID: ${BELIEFSTATE_REQUEST_ID:?Set a new UUID for this request; keep it for retries}" \
  "https://beliefstate.ai/v1/beliefs"
```

Prefer a browser? Open the [belief API playground](/api-reference/beliefs/find-investor-beliefs-or-inspect-a-full-record), select Try it, enter your key and released ticker, set limit to 1, then select Send. This makes the same metered request; it does not send automatically. Generated reference examples use schema placeholders; inspect the actual response for research data.

## Verify the result

A useful result contains a belief you can attribute and trace. HTTP 200 alone is not evidence that the response is complete or suitable for your research.

* Read `data.beliefs[0]` for the person, thesis, direction, and observation time. An empty list means no matching released belief. Check the requested ticker and filters; report the coverage gap if no relevant records are available.
* Check `data_status.mode`, `data_status.freshness`, `data_status.coverage`, and `warnings`. Preview data is for testing; preserve any stale, partial, or insufficient coverage labels.
* Follow the belief's `detail_url` with the same Bearer header for full claims, revisions, and citations. This is another metered read. Keep citation IDs attached to the claims they support.
* Treat missing confidence, horizon, or outcomes as unknown. Save `as_of` with the response so later research retains its observation cutoff.
* [Inspect belief fields and detail responses](/api-reference/beliefs/find-investor-beliefs-or-inspect-a-full-record)
* [Understand evidence and provenance](/concepts/provenance)

## Next steps

Recover from the first failed request, then expand your query.

* `401`: check the Bearer header and use an active key from [API keys](https://beliefstate.ai/account#api-keys).
* `402`: inspect [billing and available credit](https://beliefstate.ai/billing) for the account that owns the key, including spend limits.
* `404` or `503`: check [released coverage](/api-reference/tickers/find-tickers-with-available-research). The ticker, record, or service may be unavailable; don't treat this as missing credit.
* `429` or a timeout: wait before retrying. Follow Retry-After when present and reuse the same `X-Request-ID` only for the same logical request.
* [Get a research brief](/api-reference/brief/get-an-evidence-backed-research-brief-for-a-ticker)
* [Read results at a historical cutoff](/guides/point-in-time)
* [Continue through paginated results](/guides/pagination)

## Agent feedback

Report unclear, stale, or incorrect documentation through [BeliefState support](https://beliefstate.ai/contact). Include this page URL and the smallest reproducible detail.

*Last updated: 2026-09-26 · API version: v1 (Latest) · OpenAPI publication: 2026-09-13.*


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.