Skip to content

Latest commit

 

History

History
34 lines (24 loc) · 2.8 KB

File metadata and controls

34 lines (24 loc) · 2.8 KB

news

Provides crypto news querying, sentiment analysis, and full-text search for retail traders and AI agents monitoring market events.

PM: Jasmine

Business Context

Target users: retail traders and AI agents (unified underlying API; field design must accommodate both). Business priority: first feature module of News MCP; V1 delivers core news query capability. Expected call volume: medium frequency; AI Agent scenarios may poll news_get_latest every 30 seconds as a near-real-time monitoring approximation (native WebSocket push planned for V2 to reduce polling). Dependencies: no hard dependencies; pairs with the market module for a complete 'price + news' view. Risk control: all tools are Read-only with no fund operations. Site support: Global (EEA/US pending confirmation). API permissions: OpenAPI Read via OAuth 2.1, supporting OK-ACCESS-KEY/SIGN/TIMESTAMP headers; rate limits pending confirmation. Launch scope: CLI + MCP + Skill (all three surfaces in sync).

Tools

Name R/W Description
news_get_latest R Latest news sorted by time; pass importance=high for breaking news only
news_get_by_coin R News for specific coins (comma-separated)
news_search R Full-text keyword search with optional filters (coins/importance/sentiment)
news_get_detail R Full article content by news ID
news_get_domains R List available news source domains
news_get_coin_sentiment R Sentiment snapshot or time-series trend (pass trendPoints for trend mode)
news_get_sentiment_ranking R Coin ranking by hotness or sentiment direction
news_get_economic_calendar R Macro-economic calendar data (GDP, CPI, NFP, FOMC, etc.). Rate limit: 1 req/5s.
news_list_calendar_regions R List all valid region values accepted by news_get_economic_calendar (snake_case region codes).

Token 预算评估

预估 ~1600 tokens (8 tools × ~200)

典型 Workflow

  1. Scenario A — Agent generates daily market report: call news_get_latest (importance=high) to get today's top stories, call news_get_sentiment_ranking to summarize market mood, call news_get_coin_sentiment for major coin snapshots, compose report.
  2. Scenario B — User searches for a specific event (e.g. 'SEC ETF ruling'): call news_search with relevant keywords and optional coin/sentiment filters, present results list, call news_get_detail on selected article for full content.
  3. Scenario C — User asks about a specific coin: call news_get_by_coin (coins=BTC) to retrieve recent news, call news_get_coin_sentiment (coins=BTC) for bullish/bearish snapshot.
  4. Scenario D — User asks about macro events: call news_get_economic_calendar with optional region/importance filters to get upcoming or recently released economic data (e.g. NFP, CPI, GDP, FOMC). Combine with market data for impact analysis.