OpenNews Financial News Aggregator
Aggregates 85+ real-time crypto, equities, and macro news sources with AI ratings, trading signals, and bilingual summaries.
Positives: least privilege (curl/jq only, token via env), free no-token fallback endpoints, README proactively invites source-code security review, and clear disclaimers that politician/13F data is disclosure evidence not advice. Deducted: core function fully bound to closed third-party service ai.6551.io, token exposed as WSS query parameter (log-leak surface), data flow/quota consumption/terms unverified externally, no rollback or isolation design.
Positives: high internal consistency — parameter tables, enumerated business error statuses (ambiguous_entity etc.), 503 caching note, and degraded-mode guidance when aiRating.status != done. Deducted: no test suite or CI evidence in the provided files; error handling remains a documentation promise, so it stays below the static cap of 10.
Positives: clear scenarios and trigger intent (market news aggregation across crypto/equities/macro), bilingual summaries support Chinese, free fallback path declared. Deducted: boundaries only partly declared (Max subscription and quota for strategy features); core function depends entirely on overseas services (ai.6551.io / 6551.io) requiring external registration, with mainland-China reachability undisclosed — explicit environment-fit deduction.
Positives: well-layered docs (main flow → workflows → fallback → notes), clear MIT license, install guidance for many clients. Deducted: inconsistent versions (SKILL.md 1.0.7 vs package. 0.1.1 vs pyproject 0.1.0), package. homepage points to 0xfaskety/opennews-mcp while the assessed repo is 6551Team, unclear maintenance ownership/update path, no changelog.
Positives: free endpoints provide zero-threshold core value; output structures (links, bilingual summaries, signals) are complete and directly usable; workflow examples are reusable. Deducted: static review cannot verify the 85+ source claims or AI rating quality; the trading-signal value proposition rests on an unverified black-box service, so marginal value over alternatives is under-evidenced — below the static cap of 7.
Positives: detailed source lists, parameter schemas, and example responses that could in principle be independently reproduced. Deducted: everything is author-declared API documentation — no CI, no committed tests, no third-party execution evidence; the 85+ source count and prediction-signal quality lack external corroboration; facts and inferences (signal efficacy) are not separated — far below the static cap.
- Core function depends entirely on the closed third-party service ai.6551.io; data flow and terms are not auditable — review privacy/quota terms at 6551.io/mcp first.
- The WebSocket places the token in a URL query parameter, which may be captured by proxies/logs — rotate tokens periodically.
- AI ratings and trading signals are black-box outputs, not investment advice; strategy calls consume paid quota and require a Max subscription.
- Reachability of ai.6551.io from mainland China is undisclosed, and full functionality requires an external token registration, which may affect availability.
- Version numbers are inconsistent across files (1.0.7 / 0.1.1 / 0.1.0), and package. homepage does not match this repository — verify provenance before installing.
- No visible tests or CI evidence in the repository; this is a purely static review with no execution verification.
What does this skill do, and when should you use it?
OpenNews is a financial news aggregation MCP service powered by 6551.io, covering 85+ data sources across six engine categories: news, exchange listings, on-chain, meme sentiment, market anomalies, and AI predictions — spanning digital assets, U.S. stocks, semiconductors, AI infrastructure, supply chains, commodities, rates, and policy. Every article is AI-analyzed with a 0-100 impact score, long/short/neutral trading signal, and English/Chinese summaries. It supports keyword search, coin filtering, WebSocket real-time push, and strategy-triggered alerts. Without a token, a limited free hot-news endpoint is available.
Calls the ai.6551.io REST API: fetch the full source catalog, full-text keyword search, filter news by coin or source, and filter by AI score and signal. Subscribes via WebSocket (news_wss) to real-time pushes of news.update, news.ai_update, and strategy.triggered events. Provides finance-enhancement endpoints: company search and disambiguation (SEC/DART/KRX identifiers), company filings and earnings-call transcript text, key market events, U.S. House politician stock PTR disclosures, SEC 13F institution managers and holdings, and Blockscout-based on-chain wallet holdings and changes. All requests use a Bearer token (OPENNEWS_TOKEN); curl works directly.
- Crypto traders subscribe to real-time pushes of high-score BTC/ETH news and Hyperliquid whale-trade alerts during volatile markets
- Equity researchers pull SEC filings, research reports, or earnings-call transcripts for a company like SK hynix, disambiguating U.S. vs. Korean listings via exact identifiers
- Quant or arbitrage teams monitor funding-rate anomalies, large liquidations, exchange listing announcements, and 12 AI prediction signal types
- Policy-focused investors search official statements from the Fed, ECB, and U.S. Treasury channels
- Users of 'politician trading' strategies query House PTR disclosures for buy/sell records of a specific ticker
- Token-less users browse curated hot news and trending tweets by category via the free endpoints
What are this skill's strengths and limitations?
- Broad coverage: 55 news sources (Bloomberg, Reuters, FT, CNBC, CoinDesk, etc.) plus listing, on-chain, market anomaly, and prediction engines — 85+ total
- Every article ships with an AI impact score, trading signal, and per-coin ratings, enabling high-signal-density filtering
- WebSocket real-time push with precise per-coin and per-engine filters; strategy events auto-delivered to owners
- Finance-enhancement module provides verifiable evidence: SEC/DART filings, research reports, transcripts, 13F holdings, politician PTR disclosures, and on-chain wallet data
- Free token-less fallback endpoints plus a copy-paste AI security-review prompt before install
- MIT-licensed, works with a dozen mainstream MCP clients, docs in English/Chinese/Japanese/Korean
- Full search and real-time subscription require a 6551.io token; the free tier only offers curated hot news with limited search
- The token is a quota-based service — each strategy query consumes 1 quota — and pricing and rate-limit details are not documented
- Politician stock data covers only U.S. House PTR (Senate eFD explicitly unsupported); 13F holdings are delayed quarter-end filings, not real-time positions
- Some AI ratings may be pending (check status == "done"); free data is cached and returns 503 while regenerating
- No test suite shown in the repo; accuracy and latency of third-party feeds (Hyperliquid, Blockscout) are not independently verified
How do you install this skill?
- Get an API token at https://6551.io/mcp. 2a. OpenClaw: export OPENNEWS_TOKEN="<your-token>" && cp -r openclaw-skill/opennews ~/.openclaw/skills/. 2b. MCP (Claude Code): claude mcp add opennews -e OPENNEWS_TOKEN=<your-token> -- uv --directory /path/to/opennews-mcp run opennews-mcp. JSON/YAML configs for Claude Desktop, Cursor, Windsurf, Cline, Continue.dev, Cherry Studio, and Zed are also documented.
How do you use this skill?
After install, just ask your AI assistant: 'Latest crypto news', 'Search SEC regulation news', 'BTC related news', 'Important news with AI score above 80', 'Bullish signals only', or 'Subscribe to real-time news'. For direct use, call POST /open/news_search with curl (params: q, coins, engineTypes, hasCoin, score, limit, page) or subscribe to wss://ai.6551.io/open/news_wss?token=... filtering by coins and engine types. Free fallback: GET https://ai.6551.io/open/free_hot?category=macro.
How does this skill compare with similar options?
The source names a sibling project, twitter-mcp, a Twitter/X data MCP server by the same team — a lighter alternative if you only need Twitter/X data without multi-source aggregation and AI ratings.