Gatedata Signals News

GateData Signals & News

General Rules

⚠️ STOP — Read gatedata-runtime-rules.md first.

• MCP server: GateData (production)
• Scope required: signals (Plus+ plan)
• Primary tool: events_news_query (multi-task)
• Phase-1 public tasks only (API hard-rejects everything else): content_search | event_search | social_search | sentiment

When to use this skill

Do not call (rejected / out of phase-1): news, events, ugc, auto, event_detail, announcements, ratings, insight, anomalies, market_move_evidence.

Omitted task may still auto-route server-side; always pass an explicit phase-1 task in skills so calls stay predictable.

Do not use for live prices → gatedata-market-research.
Do not use for prediction odds → gatedata-prediction-markets.
Do not use for financial statements / earnings → gatedata-fundamentals-earnings.
Do not use for analyst ratings / target prices (not in this release) → say unavailable; use consensus on earnings only for estimate consensus.
Ambiguous entity → gatedata-assets-resolve first.

Task selection guide

Headlines / article search for a ticker or topic?
  → task: content_search (query and/or ticker + market)

Calendar-style / time-bounded events?
  → task: event_search (prefer from + to; short tickers: pass listing_ids when known)

Social posts?
  → task: social_search + platform (x|twitter|youtube|all)

Sentiment score (not individual articles)?
  → task: sentiment + ticker + market

Key parameters

• ticker + market — required for sentiment. Use StandardMarket (us_equity, hk_equity, crypto, …); e.g. BTC+crypto, NVDA+us_equity
• query — free text for content_search / event_search / social_search
• from + to (ISO) — time window on source_published_at (YouTube UGC and content/social feeds). Either side may be sent alone; defaults apply when both omitted
• platform — for social_search: x | twitter (alias of x) | youtube | all | web (as advertised). youtube is social_search-only (not merged into platform=all)
• has_object_refs — YouTube list only: when true, return posts that have non-empty Serving object_refs (skip macro / unlinked). Default false
• listing_ids — preferred when chaining from market tools; helps event_search / content filters for short tickers
• signal_strength — optional content filter (all or strength token as schema allows)
• window (1d | 7d | 30d), limit, mode (standard | web for content_search), source_type, allow_stale

Response fields to surface

• Crypto content_search: top-level signal_strength and per-object object_signals when present
• YouTube social_search: video_viewpoint, channel/native metadata, source_url
• sentiment sample_refs may include author_info

object_refs chaining

1. content_search / event_search / social_search may emit asset / event refs.
2. Prefer listing_ids / object_refs from market_data_query when correlating price moves with news.
3. Combine with market_data_query for price + news context.

Caveats

• Coverage strongest for us_equity, hk_equity, crypto.
• Ratings / insight / announcements / event_detail are not public on this tool.
• For “why did price move?”: market kline/snapshot + content_search/sentiment — not market task: move.

References

• scenarios.md
• troubleshooting.md