Everything about the data, the enrichment, the API and the plans — in one place. Can't find it? Email support.
Click Get free key, enter your email, and a demo key (100 requests/day, no credit card) is delivered instantly. Paid keys are activated within a few hours of payment confirmation.
Yes — 100 requests/day, forever, no credit card. The free tier returns the full enriched schema (urgency score, political lean, topic/country tags, clustering) so you can evaluate data quality before paying. It omits full article text and source names/links — those unlock on paid tiers.
Pass your key in the X-API-Key header on any /v1/* endpoint. Requests with a missing or invalid key return 401. Example:
curl -H "X-API-Key: YOUR_KEY" https://api.newsagentdata.com/v1/feed
A JSON array of fully-enriched articles, newest first, with all metadata fields populated. You control the page size with page_size (default 50, up to 200 per call) and page through results with page — so one request goes a long way against your daily quota.
Yes. We ship a Model Context Protocol (MCP) server so AI agents (Claude, etc.) can query the feed natively — tools for live feed, search, breaking news, coverage stats and source listing. Grab it at /mcp/. The REST API is plain HTTP + JSON, so it also works from any language with an HTTP client.
No separate SDK is needed — the REST API is a plain HTTP call. In Python:
import requests
r = requests.get(
"https://api.newsagentdata.com/v1/feed",
headers={"X-API-Key": "your-key"},
params={"language": "en", "min_score": 7, "page_size": 50})
print(r.json())
Interactive Swagger docs live at api.newsagentdata.com/docs, and a guided reference with examples is at /documentation/.
Yes. All sources are public RSS feeds or public Telegram channels. We collect and structure publicly available content — the same way any news aggregator works. No paywalled, private or login-gated content is included.
Yes — you can use the structured data in your own commercial products, dashboards and models under your plan. Bulk redistribution or reselling the raw dataset to third parties requires a licensing agreement; email [email protected] and we'll set you up. When displaying article links, standard practice is to credit the original outlet.
Two streams merged into one schema: thousands of RSS feeds from major outlets worldwide, plus thousands of public Telegram channels collected directly — the Telegram coverage most web-only news APIs miss entirely.
Most news APIs focus on English-language Western outlets, an inch deep. NewsAgent Data goes deep on Russian + English — including Telegram channels with no RSS feed — under one unified schema with urgency scoring, 9-category political lean, event clustering and topic/country/audience tags already applied. It's one of the only APIs offering Russian and English together, pre-enriched.
The core focus is Russian and English, with 18+ languages auto-detected across the dataset. Every article is tagged with a language field, and you can filter by language on all feed endpoints.
Currently 3,800+ active sources spanning 119 countries (100+ ISO codes). Coverage is deepest across Russia, Ukraine, the US, EU, Middle East and Asia. These numbers update live from our stats endpoint.
Sources are monitored continuously — most new articles appear within a few minutes of publication, and high-priority feeds are polled roughly every 60 seconds. Breaking articles (urgency ≥ 7) are pushed via webhook within about 60 seconds of ingestion. The /v1/breaking endpoint always reflects the last 24 hours.
The archive holds 800k+ articles and grows by roughly 1,300 articles per hour. Access depth depends on tier: Developer/Starter (7 days), Standard (90 days), Pro/Enterprise (full archive). The dataset deepens every day.
Every article carries a country tag (100+ ISO codes), a topic tag (27 categories — ukraine, markets, crypto, cyber, ai, sanctions, defense, energy, health and more), an audience segment, and a political_lean label.
Yes. Several read-only endpoints need no key: /public/breaking (recent high-urgency articles), /public/sample (a sample by country or topic), /public/stats (live totals) and /public/sources. They're great for evaluating the shape and quality of the data before you create a free key.
Full article text is available on Standard tier and above. Free and Developer/Starter tiers return the title, summary and all enrichment metadata. Source names and links are masked on the free tier and revealed on every paid tier.
Every article gets a deterministic urgency score from 0 to 10 from a 300+ pattern engine. Score ≥ 7 means breaking. It lets you route only what crosses your threshold instead of ingesting everything. See the methodology for the full signal list.
Each source is mapped to one of 9 categories: state · official · centrist · liberal · conservative · nationalist · opposition · tabloid · neutral. This lets you compare how the same event is framed across the spectrum, programmatically.
The same story reported across multiple outlets shares a cluster_id, and cluster_size tells you how widely it has spread — a fast proxy for how big a story is. De-duplicate, or measure narrative reach, in one field.
19 fields including: title, summary, url, published time, language, country, topic, audience, political_lean, urgency_score, event_type, cluster_id, cluster_size, and source metadata (revealed on paid tiers).
Yes. A single article on US Treasury sanctions targeting Russian energy could be tagged across several topics (sanctions, energy, markets) and countries (us, ru), and match more than one audience. Country tags are mention-based, so country=ua returns articles that reference Ukraine even when published by an outlet elsewhere.
Urgency, lean and clustering are produced by deterministic, rule-based pipelines — not a black-box model — so results are consistent and explainable. The methodology is published openly, and scoring caps are tuned to suppress tabloid/propaganda noise. It's signal you can audit, not a guess.
GET /v1/feed — main feed (filter by lang, min_score, days, country, political_lean, topic, audience)GET /v1/breaking — urgency ≥ 7 articlesGET /v1/search?q= — keyword searchGET /v1/audience/{segment} — audience-filtered feedGET /v1/export/jsonl — bulk JSONL export (per-tier caps)GET /v1/stream — SSE live streamGET /v1/webhook · POST · DELETE — webhook managementGET /v1/sources — source catalog with metadataFull reference: api.newsagentdata.com/docs.
The feed endpoints take page (starts at 1) and page_size (default 50, max 200). To walk the whole result set, increment page until you get fewer than page_size rows back. For large pulls, use /v1/export/jsonl instead — it streams up to your tier's row cap in a single call.
Yes — all filters are AND-combined. For example, /v1/feed?country=ua&topic=defense&language=en&min_score=5 returns only English, Ukraine-tagged, defense articles scoring 5+. Available filters: language, country, topic, audience, political_lean, event_type, min_score/max_score, and days.
Two layers: a daily quota per tier (Free 100/day, Developer 5,000/day, Starter 10,000/day, Standard 100,000/day, Pro unlimited) and a burst limit of 60 requests/minute per key. Exceeding either returns 429.
Yes, on Standard tier and above. Register a URL with POST /v1/webhook and we push score ≥ 7 articles within ~60 seconds. Each payload is signed with an HMAC-SHA256 X-Signature header so you can verify authenticity. Webhooks auto-disable after 5 consecutive failures.
Yes — GET /v1/stream is a Server-Sent Events (SSE) stream available on Pro. For lighter integrations, webhook push (Standard+) covers most real-time needs.
Yes — GET /v1/export/jsonl returns newline-delimited JSON with per-tier row caps: Free 1k, Developer/Starter 10k, Standard 50k, Pro/Enterprise 500k rows per export. The filename encodes the filters and cap used. For training an AI model on the archive, run multiple exports with different date ranges to cover the full history.
JSON on all endpoints (newline-delimited JSON for the bulk export, text/event-stream for SSE). No XML.
200 — success401 — missing or invalid API key403 — endpoint or field not available on your tier429 — daily quota or burst limit exceededFilters that exceed your tier's history window (e.g. days) are clamped to the maximum allowed rather than rejected.
Free ($0, 100/day) · Developer ($29/mo, 5,000/day) · Starter ($99/mo, 10,000/day) · Standard ($499/mo, 100,000/day, full text + webhooks) · Pro ($1,499/mo, unlimited + SSE) · Enterprise (custom). Full breakdown on the pricing page.
One API call to any /v1/* endpoint. A single call returns up to 200 articles (the page_size), so a request goes a long way against your daily quota.
Yes. Upgrade, downgrade or cancel at any time — changes take effect immediately and billing is prorated.
Yes — toggle Annual on the pricing page. Annual plans are billed yearly and include roughly two months free (about 17% off).
Paid keys are activated within a few hours of payment confirmation. For payment options and crypto settlement, contact [email protected] or message the sales bot @NewsAgentDatabot.
Standard and Pro tiers run on always-on infrastructure with Cloudflare DDoS protection and automated failover — above 99% since launch, with median API response under 200ms. Enterprise plans include a written SLA. Live status: /status/.
Keys are hashed with SHA-256 before storage — the original key is unrecoverable from our database by design. Keep your key safe; if it leaks, revoke and reissue it.
All traffic passes through Cloudflare (DDoS protection + WAF). Per-key daily quotas and a 60-req/min burst limiter (which fails closed on error) prevent runaway usage, and the demo signup form is protected by CAPTCHA.
We only index publicly available content, but if you're a rights holder and want a source or item reviewed for removal, email [email protected] with the details and we'll handle it promptly.
We store the email you sign up with and your usage logs for billing and abuse prevention — nothing more. See the privacy policy for details.
Yes. Enterprise plans include custom source lists, dedicated audience feeds, and bulk historical export built around your stack. Email [email protected] to scope it.
We offer managed monitoring, dataset licensing, white-label deployments and agent/MCP integration. See services.newsagentdata.com.
Email [email protected], or message the bot @NewsAgentDatabot — any message reaches the team directly. Standard and Pro tiers get priority response.
Reach out at [email protected] and we'll answer — and likely add it to this page.