For any client
REST API
Eight read-only endpoints covering recent trends, trend detail, retained cleaned article bodies, search, categories, domain trending, platform benefits, and platform stats.
Trend intelligence for agent runtimes
Signal over noise for AI agents.
GenticSprint ingests public web signal, deduplicates it, analyses what survives, and publishes cleaned article bodies with real provenance — over a REST API and an MCP server that agents can call today.
Call it now
Anonymous read access — no key required.
curl -sS "https://api.genticsprint.com/v1/trends/recent?hours=24&limit=5" \
-H "Accept: application/json"- OpenAPI 3.1
- MCP server
- source lineage
- retained cleaned bodies
- identical shape on every tier
Three surfaces, one record
Pick the surface that fits your runtime
The platform publishes once. Agents, applications, and dashboards read the same persisted record through the surface that suits them.
Available now
For agents
MCP server
A streamable-HTTP MCP endpoint exposing six tools and six resources. Agents get the same filters, the same fields, and the same access rules as REST clients.
For registered developers
Developer Analytics
Key management, per-key usage, benefit analytics, and webhook delivery history — with every number labelled as measured or left blank rather than guessed.
What you get
Public web signal, turned into a record you can cite
Three things hold whichever surface you call: coverage of the public web, provenance on every record, and bodies that have already been cleaned.
Public web coverage
Trends are drawn from public publisher and community sources, deduplicated across outlets, and scored for velocity and business impact so you can filter down to the signal you need.
Provenance you can check
Every trend carries its citation list — canonical URL, publisher, author, word count, and the retrieval method that produced the body. Nothing is asserted without a source you can open yourself.
Cleaned bodies, ready to read
Article bodies are retained and served as markdown, cleaned of embedded markup and personal data before storage. A trend with nothing retained says so instead of returning a stub.
Live contract examples
The exact request, in your language
These snippets are generated from the same endpoint catalogue the API is built from, so they stay correct as the surface evolves.
- Quick start
Anonymous read access needs no key. Start with recent trends and page through with bounded pagination.
- REST reference
Every endpoint with parameters, response fields, error semantics, and runnable cURL, Python, and JavaScript.
- MCP reference
Tool schemas, resource URIs, and a JSON-RPC request you can paste into an MCP client.
REST · MCP · Python · JavaScript · cURL
Switch tabs to see the same read in each client.
curl -sS "https://api.genticsprint.com/v1/trends/recent?hours=24&limit=5" \
-H "Accept: application/json"REST
https://api.genticsprint.com
JSON over HTTPS. Every path above is open to anonymous callers.
MCP · Streamable HTTP
https://mcp.genticsprint.com/mcp
The same records as tools and resources, for agents that already speak JSON-RPC 2.0.
Detail and content reads are keyed by trend_id, which the list and search endpoints return. Discovery is always two steps: read a list, take the id from a record, then fetch that record or its retained bodies.
Measured, or explicitly unmeasured
Numbers we can stand behind
These values come from the aggregate public views. A number we have not measured stays blank here — it is never rendered as zero.
Trends analysed
Distinct analysed trends currently published to the read API.
Cleaned documents retained
Sanitised article bodies stored with real provenance — the platform’s core artifact.
Publishers
Distinct publishers represented across retained source records.
Source URLs
Distinct source URLs that contributed at least one trend.
Latest pipeline update: —
Trends last 24h: —
Trends last 7d: —
“Latest pipeline update” is the most recent trend update written by the pipeline, not a deploy timestamp.
One record, verifiable end to end
Every trend carries the canonical URL, publisher, author, word count, and the retrieval method that actually produced it. You can check the claim instead of trusting it.
Cleaned before it reaches you
Injection patterns and personal data are stripped during ingestion. You are not paying to store and re-clean raw HTML on your own side.
Degradation you can detect
Unmeasured values are null, absent bodies are flagged, and empty windows are described as empty. Nothing is padded with fabricated zeros to look healthier.
Documentation
Everything needed to integrate
Written against the running implementation, including the parts that are inconvenient to document honestly.
Quick start
Anonymous read access needs no key. Start with recent trends and page through with bounded pagination.
REST reference
Every endpoint with parameters, response fields, error semantics, and runnable cURL, Python, and JavaScript.
MCP reference
Tool schemas, resource URIs, and a JSON-RPC request you can paste into an MCP client.
Schemas and errors
The OpenAPI 3.1 document, rate-limit headers, and the exact meaning of each error status.
Plans
Anonymous to evaluate, a plan to scale
Anonymous reads stay open so you can verify the data before you commit. Starter and Professional add quota, operational tooling, and support; Enterprise adds contracted volume.
Starter
per month$66
For a single integration that needs a dependable read quota and per-key analytics.
Create an accountProfessional
per month$335
For teams running the API in production across several keys and environments.
Create an accountEnterprise
annual agreementCustom
For bulk and partner integrations that need contracted volume and named support.
Talk to usQuestions
The things integrators actually ask
- Do I need an API key to read trends?
- No. The read surface is anonymous and throttled per IP. Registration gives you a higher ceiling, per-key usage analytics, and webhooks — not different data.
- What exactly is a “retained cleaned body”?
- It is the article body after structural cleaning, embedded-markup stripping, personal-data removal, and size validation — stored exactly as the pipeline produced it. If nothing survived, the API says so explicitly rather than returning a stub.
- Why do some fields come back null instead of 0?
- Because a missing measurement and a measured zero are different facts. A value with no measurement is null in the API, and treating it as 0 silently corrupts any aggregate you build on it.
- How do I find a trend_id?
- Call GET /v1/trends/recent or GET /v1/search first and read the trend_id field from each record in the response, then pass that value to GET /v1/trends/{trend_id} for the full record or GET /v1/content/{trend_id} for the retained bodies.
- Is the MCP endpoint a different dataset?
- No. The MCP tools and resources read the same tables with the same filters as the REST endpoints. The two surfaces are contract-tested for parity.
- What happens when I hit the rate limit?
- You get a 429 with a Retry-After header. Every response also reports your resolved tier and remaining quota, so you can back off before you are rejected.
Start with a single request.
Anonymous reads are open, so there is nothing to buy before you try it. The test ground renders the exact request and shows you the real response contract — including the error shapes.