论文-研报工作流 · FinAI Research Workflow

研究主题一句话 → 收到可投稿的 LaTeX 草稿。 Describe your research topic → receive a submission-ready LaTeX draft.

Quick Start (30 秒上手)

# ── 推荐方式：PyPI wheel 安装（首次约 60s）────────────────────────────
# Debian/Ubuntu：先创建虚拟环境，避免与系统 Python 冲突
python3 -m venv .venv && source .venv/bin/activate
pip install "finai-research-workflow[extras]"

# 配置 LLM（DeepSeek 直连，免费）
export DEEPSEEK_API_KEY=sk-xxxx

# 启动流水线（wheel 安装后的写作入口；加 --use-hitl 启用阶段门控）
finai-pipeline --topic "Carbon trading and green innovation" --use-hitl
# 或
python -m finai.pipeline --topic "碳排放权交易与企业绿色创新" --use-hitl

# ── 源码安装（推荐贡献者 / 想改代码的用户）────────────────────────────
git clone https://github.com/csmar432/finai-research.git && cd finai-research
pip install -e ".[extras]"
cp .env.example .env.local   # 编辑 .env.local：DEEPSEEK_API_KEY=sk-...
# 新用户：先澄清（不自动开跑）→ 再写作；或澄清时加 --continue
python scripts/start_research.py --topic "Carbon trading and green innovation"
python scripts/agent_pipeline.py --topic "Carbon trading and green innovation" --use-hitl

# ── Debian/Ubuntu apt 系统 Python ─────────────────────────────────────
# apt 的 Python 被系统管理，直接 pip install 会触发 PEP 668 冲突。
# 解法：使用虚拟环境（见上方），或加 --break-system-packages
pip install --break-system-packages "finai-research-workflow[extras]"

重要提示：缺少 DEEPSEEK_API_KEY 时，finai-pipeline 默认以退出码 4 退出（严格模式），并打印明确指引。可以用 finai-doctor 诊断配置来源。

PyPI: finai-research-workflow · 0.2.0a1 · MIT · 默认安装 pip install finai-research-workflow 不含 fastapi/streamlit（避免 PyJWT/apt 冲突） · Web 套件：pip install 'finai-research-workflow[web]' DOI: 10.5281/zenodo.21262689

Quick Demo

The demo shows an end-to-end run on a sample topic (carbon emissions trading and green innovation): tool inventory, OpenAlex literature search, empirical specification, real data acquisition from five public MCP endpoints (yfinance, SEC EDGAR, World Bank, OpenAlex, FRED), a DID coefficient table, LaTeX paper compile, and the audit_guard report. All numerical output is fetched live at demo generation time. See .github/demo/README.md for the full inventory and regeneration commands.

一次输入 → 8 阶段流水线：想法生成 → 文献综述 → 新颖性验证 → 实证设计 → 数据获取 → 分析 → 论文写作 → 对抗性 Review。每阶段需研究者确认。

3 个核心能力

⚠️ AI 生成的因果识别策略、统计结果和引用必须由研究者独立核实后方可投稿。 ⚠️ 5 个模拟数据服务器默认禁用，启用时输出带有 ⚠️ MOCK DATA 标识。

完整文档: 使用指南.md · CLAUDE.md · 运行 python scripts/setup_wizard.py --guided

Why FinAI Research Workflow?

• Built for economists, not generic AI demos — every default is calibrated for the Journal of Finance / 经济研究 standard (DID with heterogeneous treatment effects, cluster-robust SEs at the firm level, 19 robustness checks, parallel-trend plots).
• 43 MCP server directories — covers A-share financials, US equities, global macro (FRED/World Bank/IMF/OECD/BEA), and 400M+ academic papers (OpenAlex). The registry contains 28 no-key, 12 API-key, 0 stub, and 3 opt-in legal-risk directories; classification is maintained by scripts/count_assets.py.
• 58 econometric method modules, not just OLS — standard DID, event study, Bacon decomposition, staggered DID (Callaway-Sant'Anna/Sun-Abraham/Borusyak/Goodman-Bacon, requires pip install diff-in-diff2), synthetic control, instrumental variables (requires linearmodels), panel GMM, RDD, event studies, mediation, and more. See CLAUDE.md for the full list with dependency notes.
• 30 journal templates, English/Chinese/Japanese/German — JF, JFE, RFS, JAE, Econometrica, 经济研究, 金融研究, 管理世界, 会计研究, 中国工业经济.
• 18 specialised AI skills (Claude Code / Cursor / GitHub Copilot) — idea discovery, literature review, novelty check, experiment design, data acquisition, paper drafting, figure generation, LaTeX compilation, review loops.
• Human-in-the-loop, never autonomous fabrication — every stage requires explicit checkpoint approval; data sources are verified before use; no synthetic data without user consent.

Why Not Just Use ChatGPT?

FinAI is purpose-built for economic & financial research. Here is what it does that general LLMs cannot:

[!TIP] Start now with zero setup: Open in GitHub Codespaces. No local install required.

For Chinese users: The most comprehensive guide is 使用指南.md — a complete 13-chapter manual covering installation, workflows, data sources, econometric methods, paper writing, and FAQ.

Who Is This For?

Not sure? If you've ever spent days downloading data, running regressions, formatting LaTeX tables, or searching for related work — this tool is for you.

MCP Server Profile: Pick What Fits You

register_mcp_servers.py supports 4 user-type profiles — pick the one matching your hardware and use case:

# 1) Dry-run first (推荐先看)
python scripts/register_mcp_servers.py --profile academic --prune --dry-run

# 2) Actually apply
python scripts/register_mcp_servers.py --profile academic --prune

# 3) List current registration
python scripts/register_mcp_servers.py --list

See config/mcp_profiles.json for full server lists and the 使用指南.md chapter on installation for step-by-step.

Default behavior: without --profile, all 43 MCP servers are registered (matches full profile). Use --prune to remove out-of-profile servers.

Cross-Platform Installation

The project supports macOS, Linux, and Windows with platform-specific entry points:

Choose Your Path

This project supports two entry points — pick the one that matches your workflow:

Path A: AI Agent (Recommended)

The AI agent handles the full pipeline end-to-end. No need to remember commands.

# 1) Install once
./run.sh                    # macOS / Linux
run.bat                     # Windows

# 2) Health check
python scripts/health_check.py

# 3) Start an AI Agent (Claude Code / Cursor / Codex) and describe your research:
# "帮我研究关税政策对A股出口型企业创新的影响，设计一篇发表在经济研究的实证论文"

The AI agent automatically calls all 8 pipeline stages, MCP data sources, and LaTeX generators. Each stage requires your checkpoint approval before proceeding.

Path B: CLI (Script-Level Control)

Run individual scripts directly for fine-grained control:

# Full research pipeline
python scripts/agent_pipeline.py --topic "Carbon trading and green innovation"

# Research execution layer (DID/IV/RDD + writing)
python scripts/research_framework/pipeline.py --topic "Carbon trading and green innovation"

# Demo: institutional-grade financial report
python scripts/demo_research_report.py --stock 000001.SZ

# MCP tool discovery
python scripts/core/mcp_tool_market.py --search "gdp" --report

# Journal template generation
python scripts/journal_template.py --list
python scripts/journal_template.py --generate JFE output/paper.tex

Platform-Specific Notes

• macOS: Keychain is native; keyring uses KeychainBackend automatically
• Linux: Keyring uses SecretService (gnome-keyring). For Chinese fonts, install fonts-noto-cjk:

  sudo apt install fonts-noto-cjk fonts-wqy-zenhei
  

• Windows: Keyring uses Credential Manager. Chinese fonts (SimHei, Microsoft YaHei) come pre-installed

What Works Cross-Platform

• ✅ All scripts/*.py entry points
• ✅ 43 MCP servers (pure Python stdlib)
• ✅ Checkpoint (fcntl.flock falls back to no-op on Windows)
• ✅ Unit tests (pytest --collect-only; CI matrix: Ubuntu + macOS + Windows; daemon mode uses polling loop on Windows)

Known Cross-Platform Limitations

• ⚠️ event_monitor.py uses signal.pause() which is Unix-only; on Windows it falls back to a polling loop
• ⚠️ keychain_setup.py is macOS-specific; for Windows/Linux, use the cross-platform keyring via scripts/keychain_manager.py
• ⚠️ core/sandbox.py uses os.fork (Unix-only); falls back to subprocess on Windows
• ⚠️ event_monitor.py --daemon uses os.fork; on Windows the daemon exits with a friendly "use --interval 300 instead" message at startup (T2 audit 2026-07-12)
• ✅ Skills sync: knowledge/skills/, .claude/skills/, and .github/skills/ are kept in sync via python scripts/sync_skills.py (no symlinks, Windows-safe). Run after editing any skill doc.
• ✅ Codex support: AGENTS.md at root + .agents/skills/<name>/SKILL.md per skill (frontmatter with name and description). Both synced by sync_skills.py.

Show Me What It Does

Describe your research in plain Chinese — the agent handles the rest:

帮我研究关税政策对A股出口型企业创新的影响，设计一篇发表在经济研究的实证论文

What the agent produces automatically:

Footnote on numbers: The table above describes the core pipeline output stages. Idea generation, novelty verification, and literature review are separate stages that run before or in parallel. MCP server counts include 43 registered servers; some require institutional/paid accounts (Tushare Pro, Wind, CSMAR, CEIC) while others work without API keys (yfinance, akshare, World Bank, IMF, OECD, FRED, ArXiv, NBER, OpenAlex). See dependency notes in CLAUDE.md.

Architecture overview:

Multi-agent pipeline: User Input → AI Agent → 5-Stage Research Pipeline (outline → literature → plotting → writing → refinement, with optional HITL gates at each stage) → 43 MCP Servers → 58 Econometric Modules → 20 Chart Types → LaTeX Paper

Note: Demo assets are in .github/demo/ and docs/assets/. The project is actively maintained.

Contributor Setup

Source checkout

# 1. Clone the repository
git clone https://github.com/csmar432/finai-research.git
cd finai-research

# 2. Install the package with all common optional integrations
python3 -m venv .venv && source .venv/bin/activate
pip install -e ".[extras]"
# extras includes Tushare, akshare, yfinance, MCP, dashboard, and document-processing packages.

# Optional: install econometrics packages (linearmodels, pandas-datareader, pandasql, honestdid)
pip install -e ".[econometrics]"

# 3. Configure API key (at least one required)
cp .env.example .env
# Edit .env and add: DEEPSEEK_API_KEY=sk-your-key
# Other supported: ANTHROPIC_API_KEY, OPENAI_API_KEY

# 4. Run your first research pipeline
python scripts/research_framework/pipeline.py --topic "碳排放权交易对企业绿色创新的影响"

# Or use an AI Agent (recommended) for the full interactive workflow

Key numbers (auto-generated by scripts/count_assets.py):

Run python scripts/count_assets.py to regenerate these numbers. They are checked into README as a snapshot of the latest count; CI is the source of truth.

MCP Tools Overview

43 servers total: 28 work without API keys, 12 require API keys, 3 are opt-in legal-risk. See MCP Tool Marketplace for the complete catalog.

Badge	Meaning
💰 Paid	Requires institutional/paid account (Tushare Pro / Wind / CSMAR / CEIC)
⚠️ Limited	Free tier available but rate-limited or requires registration
✅ Free	No account required — works out of the box

A-share users without institutional accounts: user-yfinance (US/ADR) and user-financial (akshare free tier) cover basic equity/macro needs. Paid A-share data (CSMAR/Wind/Tushare Pro) requires institutional accounts.

See MCP Tool Marketplace Tutorial for the complete catalog.

Available Skills (18)

Each skill is documented in .claude/skills/ (Claude Code) and .github/skills/ (GitHub Copilot). In Cursor, use the Skill: command directly.

Tutorials

Documentation

Common Commands

# Paper pipeline
python scripts/research_framework/pipeline.py --topic "碳排放权交易对企业绿色创新的影响"

# Financial report
python scripts/demo_research_report.py --stock 000001.SZ

# MCP tool marketplace
python scripts/core/mcp_tool_market.py --search "gdp" --report

# Event monitor
python scripts/event_monitor.py --interval 300 --test

# Literature review
python scripts/research_framework/pipeline.py --mode lit-review --topic "carbon trading innovation"

# Or use an AI Agent directly
# "帮我做碳交易创新领域的文献综述"

# Journal template
python scripts/journal_template.py --list
python scripts/journal_template.py --generate JFE output/paper.tex

# Dashboard
streamlit run scripts/dashboard.py --server.port 8050

Data Coverage

Extending the System

Adding a New MCP Server

1. Create directory: mcp_servers/user_your_server/
2. Add SERVER_METADATA.json
3. Add tool definitions in tools/*.json
4. Register in Cursor MCP settings
5. Rebuild registry: python scripts/core/mcp_tool_market.py --dir mcp_servers

See MCP Marketplace Tutorial for full guide.

Adding a New Research Direction

1. Create file: scripts/research_directions/carbon_economics.py (copy from an existing direction like green_finance.py as template)
2. Define ResearchDirection class with:
   • Research questions
   • Data requirements
   • Hypothesis derivation
   • Empirical strategy
3. Add to scripts/research_directions/__init__.py

Contributing

Contributions welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open a Pull Request

See CONTRIBUTING.md for full guidelines.

License

This project is licensed under the MIT License. See LICENSE for details.

Acknowledgments

• 5 轮交互式澄清模式参考 Night Owl Research Agent 设计（2026-06-27 命名已重命名）
• Inspired by PaperOrchestra multi-agent architecture
• Data powered by akshare, yfinance, World Bank API, and Tushare Pro

Star History

Built With

Architecture Diagrams

Pipeline DAG (8 Stages + 4 Human-in-the-Loop Checkpoints)

flowchart TD
    Start([User inputs research topic]) --> P1[1. Outline<br/>Research framework + venue template]
    P1 -->|HITL gate| P2[2. Literature Review<br/>OpenAlex + ArXiv + Context7 + NBER]
    P2 -->|HITL gate| P3[3. Plotting<br/>Parallel chart generation]
    P3 --> P4[4. Paper Writing<br/>Full manuscript draft]
    P4 -->|HITL gate| P5[5. Refinement<br/>Multi-round adversarial review]
    P5 --> Done([LaTeX manuscript draft])

    IdeaStage[Idea Generation<br/>Stage 1 of 3] -.->|optional| P1
    DataStage[Data Acquisition<br/>Stage 2 of 3] -.->|feeds into| P3
    NoveltyStage[Novelty Check<br/>Stage 3 of 3] -.->|feeds into| P1

    style P1 fill:#e94560,color:#fff
    style P2 fill:#0f3460,color:#fff
    style P3 fill:#533483,color:#fff
    style P4 fill:#16213e,color:#fff
    style P5 fill:#0f3460,color:#fff
    style IdeaStage fill:#444,color:#fff,stroke-dasharray:3
    style DataStage fill:#444,color:#fff,stroke-dasharray:3
    style NoveltyStage fill:#444,color:#fff,stroke-dasharray:3

    classDef hitl_gate stroke:#f0ad4e,stroke-width:3px
    class P1,P2,P4,P5 hitl_gate

Pipeline stages note: The core pipeline has 5 stages (outline → literature → plotting → writing → refinement) with optional HITL gates. Idea generation, novelty verification, and data acquisition run as parallel/prior stages. The research framework CLI (scripts/research_framework/pipeline.py) provides a focused DID/IV/RDD analysis mode.

MCP Data Source Selection (43 directories: 28 no-key + 12 API-key + 3 opt-in legal)

flowchart LR
    Req[Data Request<br/>e.g. A-share ROA] --> Router{Smart Router}
    Router --> Tier1[Tier 1<br/>CSMAR/Wind<br/>Highest quality]
    Router -->|unavailable| Tier2[Tier 2<br/>Tushare<br/>+ patent data]
    Router -->|no key| Tier3[Tier 3<br/>akshare<br/>Free, slower]
    Router -->|no data| Tier4[Tier 4<br/>yfinance/synthetic<br/>Last resort]
    Tier1 --> Cache[(Local Cache<br/>SQLite)]
    Tier2 --> Cache
    Tier3 --> Cache
    Tier4 --> Cache
    Cache --> Result[Validated Data +<br/>Provenance Hash]

    style Tier1 fill:#28c840,color:#fff
    style Tier2 fill:#febc2e,color:#000
    style Tier3 fill:#0f3460,color:#fff
    style Tier4 fill:#e94560,color:#fff
    style Cache fill:#1a1a2e,color:#fff

Modern DID Estimator Selection

flowchart TD
    DID[DID with Staggered Treatment] --> Check{Never-treated<br/>available?}
    Check -->|Yes| Q1{Heterogeneous<br/>effects suspected?}
    Check -->|No| Q2{Continuous<br/>treatment?}
    Q1 -->|Yes| CS[Callaway-Sant'Anna<br/>2021 - default]
    Q1 -->|No| SA[Sun-Abraham<br/>2021]
    Q1 -->|Wants imputation| BJJ[Borusyak-Jaravel-Spiess<br/>2024]
    Q2 -->|Yes| ContDID[Continuous DID<br/>Callaway-DiTraglia 2024]
    Q2 -->|No| Decompose[Bacon Decomposition<br/>diagnose TWFE bias]
    CS --> Synth[Synthetic DiD<br/>Arkhangelsky 2021]
    SA --> Synth
    BJJ --> Synth

    style CS fill:#e94560,color:#fff
    style SA fill:#0f3460,color:#fff
    style BJJ fill:#533483,color:#fff
    style Synth fill:#16213e,color:#fff
    style Decompose fill:#16213e,color:#fff
    style ContDID fill:#0f3460,color:#fff

How FinAI Fits in the Ecosystem

FinAI focuses on the end-to-end workflow of empirical economic and finance research: research idea → literature review → empirical design → data acquisition → analysis → paper draft → submission.

General causal-inference libraries (e.g. dowhy, StatsPAI, diff-diff) focus on the CI algorithm layer. FinAI focuses on the research workflow layer that wraps data, econometrics, journal templates, and human-in-the-loop gates into one pipeline.

This focus brings complementary features for economists:

• 43 MCP data sources for A-share financials (Tushare/CSMAR/Wind), US equities (yfinance), global macro (FRED/World Bank/IMF/OECD/BEA), and 400M+ academic papers (OpenAlex/ArXiv).
• 58 econometric method modules including modern staggered DID (Callaway-Sant'Anna, Sun-Abraham, Borusyak), synthetic control/DiD, IV/2SLS, panel GMM, RDD, triple-diff, panel quantile, spatial regression, etc.
• 30 journal templates (EN+ZH+JP+DE) covering JF / JFE / RFS / JPE / Econometrica / 经济研究 / 金融研究 / 管理世界 / 会计研究 / ZWiSt / JNS and more.
• Human-in-the-loop gates at every pipeline stage to prevent LLM hallucinations.

See Related Projects below for tools that work alongside FinAI.

Maintainer

This project is maintained by @csmar432.

• 🐛 Bug reports & feature requests: GitHub Issues
• 💬 Questions & ideas: GitHub Discussions
• 🔒 Security disclosures: GitHub Security Advisories
• 💖 Sponsor / support: GitHub Sponsors · 爱发电

Contributions of all sizes are welcome — see CONTRIBUTING.md for the workflow.

Cite This Work

If this project helps your research, give it a ⭐ — it tells other economists the project is worth their time.

If you use FinAI Research Workflow in published research, please cite it as:

@software{finai2026,
  title  = {FinAI Research Workflow: An End-to-End AI Agent Pipeline for Economic and Financial Research},
  author = {csmar432},
  year   = {2026},
  month  = jun,
  url    = {https://github.com/csmar432/finai-research},
  note   = {GitHub repository. For a permanent DOI, publish on Zenodo and update this field.}
}

Related Projects

• dowhy — causal inference library (8.2K ⭐)
• StatsPAI — agent-native causal inference toolkit (274 ⭐)
• moderndid — GPU-accelerated modern DiD (25 ⭐)
• diff-diff — sklearn-like DiD in Python (280 ⭐)
• PaperOrchestra — Google's multi-agent paper writing (82 ⭐)
• E2ER-project — end-to-end empirical research pipeline (1 ⭐)
• econ-paper-studio — agent-native CLI for empirical economics (2 ⭐)

MIT License — see LICENSE for the full text.