Documentation
Two endpoints sets, one dataset. REST for any client, MCP for agent runtimes. Both are anonymous by default and both return identical fields for identical filters.
Quick start
Every response carries a schema_version, and every list response is a bounded page. There is nothing to authenticate and nothing to install.
No key is required for the anonymous read surface. Start with recent trends and page through with `limit` and `page`.
/v1/trends/recent?hours=24&limit=5
curl -sS "https://api.genticsprint.com/v1/trends/recent?hours=24&limit=5" \
-H "Accept: application/json"`GET /v1/content/{trend_id}` returns the sanitised markdown the pipeline actually stored. A trend with nothing retained returns `documents: []` and `content_retained: false` — never invented content.
/v1/content/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10
curl -sS "https://api.genticsprint.com/v1/content/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10" \
-H "Accept: application/json"Search accepts the same category, time-window, impact, and quality filters as the recent-trends surface, so a query and a browse are the same shape.
/v1/search?q=on-device%20inference&limit=10
curl -sS "https://api.genticsprint.com/v1/search?q=on-device%20inference&limit=10" \
-H "Accept: application/json"Trend discovery
Detail and content endpoints are keyed by trend_id. Discovery is always a two-step flow: page or search the trend pool first, read trend_id from a record, then fetch the full record or its retained bodies.
GET /v1/trends/recent?hours=24&limit=5
Or filter by category, velocity, min_impact, and page. Each record in the trends array carries trend_id.
GET /v1/search?q=on-device%20inference&limit=10
The alternative entry point when you already know what you are looking for. Records carry the same trend_id.
GET /v1/trends/{trend_id}
Full record plus source lineage: citations, canonical URLs, publishers, and the real fetch method.
GET /v1/content/{trend_id}
Retained cleaned bodies for the same id, in markdown, with content_retained reporting whether anything survived.
Every trend_id is a stable UUID. The same value works on REST and on MCP (`get_trend_detail` / `get_trend_content`), so an agent can discover over one transport and fetch over the other.
REST API
All eight are GET, all eight are anonymous, and all eight return the same schema_version. Any filter accepted by the browse surface is accepted by search.
| Method | Path | Parameters | Returns |
|---|---|---|---|
| GET | /v1/trends/recent | category, hours, velocity, min_impact, min_quality, page, limit | Filtered, paginated trend summaries ordered by business impact then recency. |
| GET | /v1/trends/{trend_id} | trend_id (path) | Full trend record plus source lineage: canonical URL, publisher, author, word count, real fetch method. |
| GET | /v1/content/{trend_id} | trend_id (path), limit | The sanitised, PII-redacted article bodies exactly as the fetch stage produced them. Same shape for anonymous and registered agents. |
| GET | /v1/search | q (required), category, hours, limit | Keyword search across the trend pool with the same filters as the recent-trends surface. |
| GET | /v1/categories | none | Every category with trend counts and average impact scores. |
| GET | /v1/sources/{domain}/trending | domain (path), limit | Trends originating from one publisher domain, with that domain’s quality score. |
| GET | /v1/benefits | none | Anonymous aggregate benefit metrics. Unmeasured values are null, never a fabricated zero. |
| GET | /v1/stats | none | Total trends, freshness, and source diversity for the platform. |
Each block below uses the real sample path, so the snippet is one you can run unchanged against the base URL above.
Filtered, paginated trend summaries ordered by business impact then recency.
parameters: category, hours, velocity, min_impact, min_quality, page, limit · sample: /v1/trends/recent?hours=24&category=tech&limit=5
curl -sS "https://api.genticsprint.com/v1/trends/recent?hours=24&category=tech&limit=5" \
-H "Accept: application/json"Full trend record plus source lineage: canonical URL, publisher, author, word count, real fetch method.
parameters: trend_id (path) · sample: /v1/trends/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10
curl -sS "https://api.genticsprint.com/v1/trends/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10" \
-H "Accept: application/json"The sanitised, PII-redacted article bodies exactly as the fetch stage produced them. Same shape for anonymous and registered agents.
parameters: trend_id (path), limit · sample: /v1/content/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10
curl -sS "https://api.genticsprint.com/v1/content/8f2c0d1e-4a7b-4c9d-9f31-2b7a5c4e6d10" \
-H "Accept: application/json"Keyword search across the trend pool with the same filters as the recent-trends surface.
parameters: q (required), category, hours, limit · sample: /v1/search?q=on-device%20inference&limit=10
curl -sS "https://api.genticsprint.com/v1/search?q=on-device%20inference&limit=10" \
-H "Accept: application/json"Every category with trend counts and average impact scores.
parameters: none · sample: /v1/categories
curl -sS "https://api.genticsprint.com/v1/categories" \
-H "Accept: application/json"Trends originating from one publisher domain, with that domain’s quality score.
parameters: domain (path), limit · sample: /v1/sources/techcrunch.com/trending?limit=10
curl -sS "https://api.genticsprint.com/v1/sources/techcrunch.com/trending?limit=10" \
-H "Accept: application/json"Anonymous aggregate benefit metrics. Unmeasured values are null, never a fabricated zero.
parameters: none · sample: /v1/benefits
curl -sS "https://api.genticsprint.com/v1/benefits" \
-H "Accept: application/json"Total trends, freshness, and source diversity for the platform.
parameters: none · sample: /v1/stats
curl -sS "https://api.genticsprint.com/v1/stats" \
-H "Accept: application/json"MCP server
MCP clients call tools with JSON arguments and read resources by URI. The filters, the fields, and the access rules are the same as the REST surface — the surfaces are contract-tested for parity.
| Tool | Arguments | Required |
|---|---|---|
| get_recent_trends | { category, hours, velocity, min_impact, min_quality, page, limit } | — |
| search_trends | { query, category, time_window, min_impact, min_quality, limit } | query |
| get_trend_detail | { trend_id } | trend_id |
| get_trend_content | { trend_id, limit } | trend_id |
| list_categories | {} | — |
| get_source_trends | { domain, limit } | domain |
Six URI-addressed reads for clients that prefer resources to tool calls.
The MCP tab shows the exact tools/call envelope.
curl -sS "https://api.genticsprint.com/v1/trends/recent?hours=24&category=tech&limit=5" \
-H "Accept: application/json"Access and auth
An unregistered caller receives the same fields, the same provenance, and the same retained bodies as a registered one. Registration changes quota and operational tooling, never the content shape.
| Tier | Who | Shape | Throttled |
|---|---|---|---|
| public | Anonymous callers | Full read surface: trends, detail, provenance, retained cleaned bodies, benefits, stats. | yes |
| registered | Developers with an API key | The same content shape with higher throughput ceilings, per-account usage analytics, and webhooks. Pagination bounds are identical for every tier. | yes |
| enterprise | Bulk and partner integrations | Reserved ceilings configured by environment. | yes |
Rate-limit state is on every response. Back off using the reported remaining value rather than waiting for a rejection.
| Header | Meaning |
|---|---|
| X-RateLimit-Limit | Per-minute ceiling for the caller’s resolved tier. Omitted when the tier has no ceiling. |
| X-RateLimit-Remaining | Requests left in the current minute. Omitted when the tier has no ceiling. |
| X-RateLimit-Tier | Resolved tier for this request: public, registered, or enterprise. |
| Retry-After | Seconds to wait — only present on a 429 response. |
Higher ceilings are configured per account rather than sold as a separate content shape, so a plan change never changes the fields a client reads.
Schemas
The OpenAPI 3.1 document is the normative description of every request and response. Where this page and the document disagree, the document is correct and this page is a bug.
Includes every path, parameter bound, response schema, error response, and the rate-limit header definitions. Canonical copy also published at https://www.genticsprint.com/docs/openapi.json.
| Status | Meaning |
|---|---|
| 400 | Invalid filter value — for example `hours=0` or `hours` above the 90-day cap. |
| 404 | Trend not found, or the endpoint path does not exist. |
| 429 | Tier ceiling reached. Read `Retry-After` before retrying. |
| 503 | Database temporarily unavailable. Retryable. |
A 401 or 403 is not in the table above on purpose. Those statuses are produced by the platform in front of the application, not by the API itself, so they are not part of the documented contract. Treat them as a transient availability condition and retry.
Examples
The test ground builds the same bounded requests this page documents and shows you the raw response, so you can confirm the contract against the live surface before writing any client code.
Pick an endpoint, set the filters within their published bounds, and send the request. No key required and no state stored.
Python, JavaScript, and cURL snippets appear next to every endpoint above. They are generated from the endpoint catalogue, so they stay valid as the contract evolves.
https://api.genticsprint.com
Every field on this page comes from the live read surface, and every example is runnable as written. Register when you want higher throughput ceilings, per-account usage, and webhooks — never to unlock a different content shape.