Agents Gate
Give this page to your agent — it will immediately understand what to do with it. In truth, this page is not really for humans: it is for agents, to help them make better decisions in trading and market analysis. Either way, we do not recommend using the news backdrop as the only signal for opening a position — use a combination of factors to make the highest-quality decisions.
Every number this site shows is available as JSON. Open — no key, no account, no quota. Readings for Bitcoin, Gold and Oil, refreshed hourly, plus an immediate extra run when an asset's price moves sharply.
Try it now
curl "https://positivesearch.app/api/v35/latest?asset=BTC"
Returns the current reading with its timestamp and expiry, the narratives driving it, the X (Twitter) card, and every scored headline with its own score and a link to the original article:
{
"asset": "BTC",
"index": 0.42, "blended": 0.55,
"state": "strongly_bullish",
"change_1h": -0.004, "change_24h": 0.157, "change_7d": 0.21,
"fetched_at": "2026-08-20T15:50:22Z",
"expires_at": "2026-08-20T16:50:22Z",
"is_stale": false, "trigger": "hourly",
"methodology_version": "v4-2026-07-19",
"confidence_basis": { "articles_scored": 176, "distinct_sources": 38,
"fresh_share_24h": 0.62, "cross_source_agreement": 0.81 },
"narratives": [ { "label": "Treasury buyback sparks Bitcoin surge",
"mass": 8.3, "agreement": 0.88, ... } ],
"articles": [ { "title": "...", "score": 0.7, "url": "..." }, ... ]
}
Endpoints
| Endpoint | Returns |
|---|---|
GET /api/v35/latest?asset=BTC|GOLD|OIL |
current reading, narratives with member articles, X card, per-headline scores |
GET /api/v35/sentiment?asset=BTC|GOLD|OIL |
the index as a 7-day time series — the same window the site's chart draws |
GET /api/v35/candles?asset=BTC|GOLD|OIL |
hourly price candles for the same instrument, aligned with the index |
GET /api/v35/polymarket?asset=BTC|OIL |
the prediction-market leg: events used, liquidity, volume, resulting skew |
GET /.well-known/agents-gate.json |
everything at once — current readings for all three assets plus this whole map, in one machine-readable document |
Everything else under /api/ is closed internals and answers 403.
Machine-readable spec with types and required fields:
/openapi.json (OpenAPI 3.1).
Reading a value correctly
Every response carries fetched_at (when it was measured),
expires_at (when it stops being the current reading) and
refresh_seconds. The rule is mechanical: if now > expires_at,
re-fetch — an expired value is not wrong, it is historical. Quote a value together with
its timestamp.
Two numbers per reading: index is news coverage alone;
blended folds in X chatter and Polymarket odds — it is the headline number
the site's pages show, and the one to quote. state is the same number as a
machine-readable label, on the exact thresholds the dashboard uses;
change_1h / change_24h / change_7d are precomputed
deltas (null when no run exists near that moment — never a farther point in disguise).
confidence_basis is deliberately facts, not a composite score: article count,
distinct sources, freshness share, cross-source agreement — compose your own trust.
methodology_version marks comparability: readings sharing it are computed
the same way.
What the index is — and is not
The index measures the tone of world news coverage on a −1..+1 scale, one model scoring each headline for sentiment and materiality. It does not measure price and it does not forecast price. It is one instrument among several — combine it with whatever analysis you already run.
Verify us yourself
Every reading ships its full article list — each headline with its own score, publication time and a link to the original. Past readings cannot be silently rewritten: every run enters a sha256 hash chain whose head is anchored daily in the Bitcoin blockchain via OpenTimestamps (manifest · chain heads). Methodology in plain text: /llms.txt.
For AI agents
The canonical machine entry point is /.well-known/agents-gate.json — one request returns the current readings for all three assets, honest limits, this endpoint map and how to cite. Point your agent there.
Questions or an integration to discuss: hi@positivesearch.app