cryptonews.is
Server Details
Live crypto news from 1,000+ sources, grouped into stories, in 28 languages, with news sentiment.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- OAuth
- Works in Glama
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- Repository
- cryptonews-is/mcp
- GitHub Stars
- 2
- Server Listing
- cryptonews.is MCP Server
- ChatGPT
- –
- Claude
- cryptonews.is
TDQS
Scored across 14 tools
Most tools target clearly distinct resources: market quotes (get_market), single-coin snapshot (get_coin), technical readings (get_technicals), sentiment (get_sentiment vs get_sentiment_history), story detail (get_story), and search (search_stories). The main overlaps are latest_stories vs changes_since (both return ranked story lists, distinguished only by cursor semantics) and get_coin, which bundles quotes, technicals verdict, stories and sentiment that are each also available through dedicated tools.
All 14 names are snake_case and built from a small, predictable vocabulary: ten use the get_<resource> pattern (get_story, get_coin, get_sentiment_history), with latest_stories, search_stories, changes_since and status as readable noun/verb variants. No casing or style mixing; an agent can infer the family a tool belongs to from its name.
14 tools is well within the ideal range and each one maps to a distinct product capability (stories, search, deltas, sentiment, history, market, coin, technicals, impact, calendar, developing events, brief, status). Nothing appears redundant enough to remove, and nothing trivial is included just to pad the surface.
The surface covers the full news lifecycle: discovery (latest_stories, search_stories, changes_since), depth (get_story, get_developing), analytics (sentiment, history, impact, technicals, market), context (calendar, brief) and ops (status). The one real gap is a dead end: get_coin tells the agent to 'use get_technicals or get_candles for more detail', but no get_candles tool exists, and there is no coin-listing/lookup tool to discover valid symbols beyond those surfaced in stories.
Available Tools
14 toolschanges_sinceChanges sinceARead-onlyInspect
Standard and Pro: changes since an edition cursor, ranked by prominence. IDs in removed that were re-keyed appear in replaced_by with their new id. limit defaults to 100 each for added, changed and removed; detail defaults to compact. copy_pending marks missing copy; copy_skipped: the story is rated negligible and will not get a summary (it is dropped if copy arrives later). The cursor advances; when truncated is present (added/changed/removed totals), call latest_stories for the full current list. A malformed cursor returns 400 invalid_cursor. While the upstream is unavailable an unsaved cursor returns itself with empty lists and stale:true: poll again with it. reset:true (old epoch or expired cursor) carries empty lists: re-read latest_stories and continue from the returned cursor. contribution on full-detail items is a snapshot; changes to it alone are not reported.
Inputs: cursor (required), limit, detail.
get_briefDaily briefARead-onlyInspect
Free, Standard and Pro: the daily brief of a completed UTC day (latest by default; day YYYY-MM-DD from available_from): a cited headline and paragraphs, the day's closing sentiment, lead, top and other stories, and the developing event. Each paragraph lists citations (phrase, story_id, reader_url); cite those story ids. Today's brief is written shortly after 00:00 UTC; 404 for unwritten days. English.
Inputs: day.
get_calendarComing upARead-onlyInspect
Standard and Pro: dated events coming up (unlocks, votes, hearings, launches, deadlines, macro releases, listings) found in the news, from today for 30 days by default (from/to YYYY-MM-DD, at most 92 days). status says how firm the date is (announced, reported or expected); precision is day, week, month or quarter; source_count counts the reports behind it. Cite the story_ids. Filter with kind, coin or min_sources (2+ for corroborated entries).
Inputs: to, coin, from, kind, +2 more.
get_coinCoinARead-onlyInspect
Free, Standard and Pro: one coin in one call: its quote, rank and window changes, the technicals verdict, the latest stories that mention it and the mean sentiment of those stories over 4 and 24 hours (not the market reading). Use get_technicals or get_candles for more detail.
Inputs: symbol (required), limit.
get_developingDeveloping eventsARead-onlyInspect
Standard and Pro: a developing big event (for example a major hack) as a cited synthesis. Without slug: the active and recently ended threads with stories. With slug: the cited summary (brief, figures, ledger of who said what, and the key timeline entries), each entry with up to 5 story_ids (the earliest and newest reports first; story_ids_total counts all); timeline_total counts every timeline entry. section (brief, figures, ledger, timeline, threads or drift) returns that section in full with every story id: section=timeline gives every dated entry and every source behind it. Cite the story ids listed on the entry you use (get_story resolves them); member_count counts the thread's stories. figures and drift entries keep value as written and add a numeric amount (with qualifier, e.g. 'more than') when the value is a plain number. Pro: as_of (ISO 8601 UTC, not in the future) replays what was recorded at that instant, with nothing learned later; the body carries as_of, snapshot_at (the record used, at or before as_of) and replay: recorded, or partial with not_recorded and a note naming what is not from the instant. With a slug only: the page revision published by as_of, with the story ids it cited then.
Inputs: slug, as_of, section.
get_impactPrice impactARead-onlyInspect
Standard and Pro: how a coin's price moved around past stories. Each recorded story is one real event, timed at its earliest report across outlets and languages; its coins' prices were recorded just before publication and 15 minutes, 1, 4 and 24 hours after, with BTC alongside. Give coin, or story for one story's coins. Filter with category, genre, kind (the story's calendar entry kind), direction (bullish or bearish news, from the story's sentiment score) and min_prominence (0-1); from/to (YYYY-MM-DD, both included, at most 366 days) default to the last 90 days, and Standard reaches 90 days back. summary covers every match up to 2,000 (total counts them; truncated, beside total, says the answer was cut, and scanned_from says how far back it is complete): per horizon n, median, p25, p75, up_share and median_vs_btc, in percent, with n alone when fewer than 5 samples. before_1h is the move over the hour before publication, so you can see how much of a move came first. events (limit default 20, max 100) list the stories newest first with price_at, change and vs_btc (percent from price_at; vs_btc is the coin's change minus BTC's), source, price_basis (candle: the close of a one-minute exchange candle; sample: a CoinGecko price sampled about every five minutes) and complete (false while later checkpoints are still due). Past observations with sample sizes, not a forecast. Cite stories by id or reader_url.
Inputs: to, coin, from, kind, +6 more.
get_marketGet marketARead-onlyInspect
Free, Standard and Pro: market quotes, top coins, total market cap and the coins named in current stories. symbols filters quotes, top coins and story coins; total market cap is always included; quote volume_24h is in coin units, not USD.
Inputs: symbols.
get_sentimentGet sentimentARead-onlyInspect
Free, Standard and Pro: current sentiment and its strongest drivers (limit, default 20, max 50; Free: at most 20, and a larger limit adds limit_applied:20; driver_count gives the total). The drivers list stories with a summary first, strongest first within each group. detail defaults to compact (~550 bytes per driver); full keeps all story fields. copy_pending marks missing copy; copy_skipped: the story is rated negligible and will not get a summary (it is dropped if copy arrives later); language null means the copy is not written yet. Pro: as_of (ISO 8601 UTC, not in the future) replays what was recorded at that instant, with nothing learned later; the body carries as_of, snapshot_at (the record used, at or before as_of) and replay: recorded, or partial with not_recorded and a note naming what is not from the instant. The replayed reading is the minute's, from 2026-10-29T02:17Z; its drivers are the story list recorded every 15 minutes.
Inputs: as_of, limit, detail, window.
get_sentiment_historySentiment historyARead-onlyInspect
Standard and Pro: sentiment history (Standard: 30 days back). from/to may span at most 72 hours per call (400 invalid_arguments otherwise); from defaults to 24 hours before to; page through longer spans. recorded_from: minute history is recorded from 2026-10-29T02:17Z; earlier ranges return only the native’s live history or nothing. resolution defaults to auto (target 300 points); raw keeps all; buckets use the last sample score at the bucket start. Returns resolved resolution and available_from, the earliest returned point or null.
Inputs: to, from, window, resolution.
get_storyGet storyARead-onlyInspect
Standard and Pro: story detail and publisher reports, including archived stories. A story’s id can change once grouping settles, and old ids keep resolving to the current story: a merged id returns the current story with replaced_by (its id) and requested_id (the id you asked for); archived marks archived responses. During an upstream restart the story may come from a saved copy with stale:true and age. Story language identifies the copy (null means the copy is not written yet); copy_pending marks missing copy; copy_skipped: the story is rated negligible and will not get a summary (it is dropped if copy arrives later). The first 8 reports each carry the outlet’s own English headline and summary (null when none; report_copy.with_copy counts them). An id prefix of at least 8 characters resolves when unique (else 400 ambiguous_id with candidates). impact, when present, lists the prices recorded around the story for each of its coins (as get_impact events). Cite stories by their full id or reader_url; ids are stable, and merged ids return replaced_by. Pro: as_of (ISO 8601 UTC, not in the future) replays what was recorded at that instant, with nothing learned later; the body carries as_of, snapshot_at (the record used, at or before as_of) and replay: recorded, or partial with not_recorded and a note naming what is not from the instant. A replayed story has its recorded text and values; reports are never recorded, so they are the current reports published by as_of and not_recorded lists them.
Inputs: id (required), as_of, language.
get_technicalsGet technicalsARead-onlyInspect
Free, Standard and Pro: technical readings (trend, momentum, stretch) for ten large coins over today, this week and this month, with the signals behind them on Standard and Pro. timeframe requires symbol; symbol+timeframe returns that one reading. detail defaults to compact; full adds the score history and, with timeframe, the candle series. Signals, not advice.
Inputs: detail, symbol, timeframe.
latest_storiesLatest storiesARead-onlyInspect
Free, Standard and Pro: ranked current stories (limit default 20, max 100; Free: at most 20, and a larger limit adds limit_applied:20). Categories: markets, economy, regulation, technology, security, business, adoption. copy_pending marks missing copy; copy_skipped: the story is rated negligible and will not get a summary (it is dropped if copy arrives later); language null means the copy is not written yet. Pro: as_of (ISO 8601 UTC, not in the future) replays what was recorded at that instant, with nothing learned later; the body carries as_of, snapshot_at (the record used, at or before as_of) and replay: recorded, or partial with not_recorded and a note naming what is not from the instant. Replayed lists are recorded every 15 minutes, carry no summaries (get_story with as_of has each story's text then) and are English.
Inputs: coin, as_of, limit, window, +2 more.
search_storiesSearch storiesARead-onlyInspect
Pro: full-text story search (limit default 20, max 50; offset 0–1000, default 0). match defaults to auto: stories containing every term, or any term when none contain them all; the response's match says which was used. Query terms use English stemming (English stop words ignored; acronyms such as SEC, ETF or DPRK match their spelled-out forms, and a capitalised coin symbol matches the coin name). Ranked by relevance: more matched terms, especially in the headline, then boosts for live, summarised and recently updated stories; recent developments (updated_at, newest report) rank above older first reports of comparable relevance. since/until filter on event_at (first report). Twins of one article (same publisher url or headline) are collapsed and total counts them once. Categories: markets, economy, regulation, technology, security, business, adoption. Results mark archived:false for current stories and archived:true for archived ones (left out if the current edition is unavailable); archived ids resolve via get_story, which may return replaced_by; merged ids are returned as their current story with replaced_by. Each result’s reports (up to 3: the earliest published report, usually the original source, then those matching the query; summaries cut to two sentences; reports_total counts all of the story’s reports, and get_story lists them whole) carry an outlet’s own headline and summary. A result in a developing event carries developing {slug,label,state,url}: call get_developing with that slug for its timeline, mechanism and who said what, and cite the story ids listed on the entry. Cite every sentence that states news, the opening one included; cite the story whose own text, or one of its listed reports, carries the detail (for a report, also name source_name and published_at); do not add a generic "Sources: cryptonews.is" line. copy_pending marks missing copy; copy_skipped: the story is rated negligible and will not get a summary (it is dropped if copy arrives later); language null means the copy is not written yet. Cite stories by their full id or reader_url; ids are stable, and merged ids return replaced_by.
Inputs: query (required), coin, limit, match, +4 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Multi-language crypto news with editorial sentiment + importance scoring; cites original publisher.
Real-time crypto and stock news scored by market impact, tagged with the assets each story moves.
Deduplicated crypto headlines from 37 RSS sources. Free preview; $0.0008/call via x402.
Real-time curated crypto news for AI agents with sentiment, recaps, and search.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceReal-time crypto news aggregation with sentiment scores and token mentions, enabling market analysis and trading signals via pay-per-call micropayments.MIT
- FlicenseNot gradedqualityBmaintenanceFree, no-auth crypto news API with 200+ sources for fetching real-time and historical crypto news.135 npm331-
- AlicenseBqualityCmaintenanceRead-only crypto news and source-linked coin events with observed 5-minute, 1-hour and 24-hour price reactions from WhyCoinMoves. Includes publication/detection timestamps and explicit methodological limitations; no API key required.2MIT
- AlicenseAqualityDmaintenanceProvide the latest cryptocurrency news to AI agents.174MIT
Glama MCP Gateway
Add one secure layer between your agents and this server.