界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
开发与工程 · SKILL.md

llmquant-events

llmquant-events

【开发与工程】Router skill for LLMQuant event workflows. Use when the user needs earnings event briefs, M&A tracking, regulatory risk, catalysts, event calendars, or cross-asset event impact.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: llmquant-events description: Router skill for LLMQuant event workflows. Use when the user needs earnings event briefs, M&A tracking, regulatory risk, catalysts, event calendars, or cross-asset event impact. input_data_source: LLMQuant Data category: events

LLMQuant Events

This category routes event-driven research workflows for earnings, M&A, regulatory catalysts, and event-risk monitoring.

Routing Rules

  1. Identify event type, issuer, asset, date, jurisdiction, affected instruments, and requested output.
  2. Select the closest workflow below.
  3. Open only that workflow and any referenced local resources.
  4. Use LLMQuant Data for filings, prices, options, estimates, news, corporate actions, regulatory records, prediction markets, and macro context.
  5. Report event dates, filing dates, data timestamps, source periods, stale notices, and missing inputs.

Workflow Index

User intentWorkflow
Build an earnings-event brief with setup, expectations, options, and risk cases.workflows/earnings-event-brief.md
Track M&A, deal spread, approvals, financing, and break-risk milestones.workflows/mna-event-tracker.md
Monitor regulatory, legal, policy, antitrust, FDA, or geopolitical event risk.workflows/regulatory-risk-monitor.md

LLMQuant Data Contract

Prefer LLMQuant Data when available. The workflows may need these data capabilities:

  • Retrieve event calendars, corporate actions, filings, news, estimates, transcripts, and issuer profiles.
  • Retrieve price history, options, implied move, event-window returns, volume, volatility, and sentiment.
  • Retrieve M&A terms, deal milestones, financing, regulatory approvals, court dates, and antitrust records when available.
  • Retrieve prediction-market odds, macro releases, policy calendars, and cross-asset context when relevant.

Fallback:

  • If event-specific data is unavailable, name the missing input and avoid event-probability or spread conclusions that depend on it.
  • Do not invent dates, deal terms, legal deadlines, or regulatory decisions.
我的收藏