finance MCP Server
PBX Market Estimation & VoIP Trend Forecasting
Discovered via github-topic:mcp-server and last synced 3mo ago.
Install instructions not detected yet
Check the source repository for the latest setup steps.
Meaning
Cloudflare R2 Account API Token，例如你的 `pbx_application_token`。這是給 Wrangler remote upload 用，不是 R2 S3 Access Key ID。
Optional token required by your unofficial NotebookLM bridge, if any. Leave empty if the bridge does not require bearer auth.
Cloudflare 帳號頁面取得，或登入後執行 `npx wrangler whoami`。
Data
Countries
### Project Structure ``` pbx_estimation/ ├── data/ │ ├── raw/ # API downloads cache │ └── processed/ # Cleaned panel data ├── notebooks/ │ ├── 01_fetch_data.ipynb # Data collection │ ├── 02_eda_visualization.ipynb # Exploratory analysis │ ├── 03_logistic_growth.ipynb # S-Curve fitting │ └── 04_survival_analysis.ipynb # CoxPH model ├── src/ │ ├── data/fetcher.py │ ├── data/preprocessor.py │ ├── models/logistic_growth.py │ └── models/survival.py ├── tests/ ├── .github/workflows/report.yml # Auto-generate reports ├── config.yaml ├── requirements.txt └── pyproject.toml ``` ### Getting Started ```bash # Clone & install git clone https://github.com/dennis-lee/pbx_estimation.git cd pbx_estimation python -m venv venv && source venv/bin/activate pip install -r requirements.txt # Run notebooks jupyter notebook notebooks/ ``` ### Report Generation via GitHub Actions Reports are automatically generated and published via a scheduled GitHub Actions workflow: - **Schedule**: Runs every hour at minute 0 UTC - **Trigger**: Also supports manual dispatch via GitHub UI - **Steps**: 1. Fetch latest data from World Bank / ITU / BEREC APIs 2. Execute all Jupyter notebooks in order 3. Render notebooks to HTML and PDF 4. Archive reports as build artifacts (downloadable from Actions tab) 5. Optionally deploy to GitHub Pages for a live dashboard > 💡 **Manual trigger**: Go to `Actions` → `Report Generation` → `Run workflow` → `Run now` ### Cloud RAG Endpoint and Asset Sync The technology alternatives page uses browser-side keyword filtering for quick narrowing, then calls the Cloudflare Worker in `rag_engine/` for cloud RAG prioritization. CI builds a RAG asset manifest from `reports/`, `data/processed/`, and `frontend/data/`, uploads those assets to Cloudflare R2 when configured, and the Worker retrieves report/data evidence from that bucket. Set these repository secrets for Cloudflare RAG: - `CLOUD_RAG_ENDPOINT` - `CLOUDFLARE_API_TOKEN` - `CLOUDFLARE_ACCOUNT_ID` - `CLOUDFLARE_R2_BUCKET` (current Worker binding expects `auto-rag`) - Optional S3 API upload secrets: `CLOUDFLARE_R2_S3_ENDPOINT`, `CLOUDFLARE_R2_S3_ACCESS_KEY_ID`, `CLOUDFLARE_R2_S3_SECRET_ACCESS_KEY` Alternatively, the same self-developed RAG engine can run on Hugging Face Spaces as a Docker Space using the files in `rag_engine/`. Set `HF_TOKEN` and `HF_SPACE_ID` to let CI upload the Docker Space. Hugging Face's default CPU Basic Space is currently free and provides 2 vCPU, 16 GB RAM, and 50 GB non-persistent disk. The Space endpoint can also be used as `CLOUD_RAG_ENDPOINT`. The endpoint returns: ```json { "recommendation": "short explanation", "alternatives": [{ "name": "MQTT (MQTT-SN)", "rank": 1, "reason": "why it fits" }], "solutions": [{ "name": "Twilio Programmable Voice", "rank": 1, "reason": "why it fits" }], "documents": [{ "name": "reports/global_research_report_zh.md", "rank": 1, "excerpt": "retrieved report/data evidence" }] } ``` ### NotebookLM Google NotebookLM does not have an official public file-upload API. CI therefore creates a NotebookLM-ready bundle at `rag_engine/dist/notebooklm_sources/` and includes it in the workflow artifact for manual upload. If you run an unofficial bridge such as `notebooklm-rest-api` or `notebooklm-py`, set `NOTEBOOKLM_UPLOAD_URL` and optionally `NOTEBOOKLM_API_TOKEN`; CI will post the selected sources to that endpoint. ### How to Fill `.env.example` Copy `.env.example` to `.env` for local work, and add the same names as GitHub repository secrets when CI needs them.
Secret Access Key shown once when creating the R2 S3 API token. Store only as a local `.env` value or GitHub secret.
Public RAG endpoint used by the frontend build. Use either your Cloudflare Worker URL, for example `https://pbx-rag-engine.<account>.workers.dev/`, or your Hugging Face Space URL, for example `https://<user>-pbx-rag-engine.hf.space/`.
Same endpoint as above, stored as a GitHub secret so `.github/workflows/report.yml` can pass it into `NEXT_PUBLIC_CLOUD_RAG_ENDPOINT` during the Pages build.
通常維持 `latest`。只有要分 staging/date namespace 時才需要修改。
Optional. Only set this if you run an unofficial NotebookLM bridge such as `notebooklm-rest-api` or `notebooklm-py`. Use that bridge's upload endpoint URL. Leave empty for manual NotebookLM upload.
NotebookLM bridge 上傳 timeout 秒數，預設 `60`。
Cloudflare R2 建立 bucket；目前預設 `auto-rag`。
Runtime port for the Hugging Face Docker Space/local server. Hugging Face Docker Spaces should use `7860`.
選填的 S3 API 路徑。如果只使用 `pbx_application_token`，請讓所有 S3 欄位保持空白。若使用 S3 API 模式，你的 bucket 可填 `https://8dfc8c4994bd0925c72ab9e2eff79b48.r2.cloudflarestorage.com/auto-rag`。
Runtime host bind address. Use `0.0.0.0` for Docker/Hugging Face Spaces.
Cloudflare Dashboard → R2 → Manage R2 API Tokens → create an R2 token with object read/write access for the `auto-rag` bucket. Use the 32-character Access Key ID. Do not use the Cloudflare API token or token id here.
RAG engine CORS 設定；測試可用 `*`，正式環境建議填 GitHub Pages origin。
Hugging Face Settings → Access Tokens 建立具 Spaces 寫入權限的 token。
Cloudflare Worker 是否使用 Workers AI 產生摘要；Hugging Face Docker Space 使用 deterministic ranking。
Hugging Face Space repo id，格式 `<username-or-org>/<space-name>`。
Hugging Face Docker container 內的資產路徑，除非改 Dockerfile，維持 `/app/dist/hf_assets`。
Connect to Helium's MCP server for news research, media bias analysis, balanced perspectives, stock/options data, and semantic meme search across 3.2M+ articles and 5,000+ sources
Automate account reconciliation by matching transactions, identifying discrepancies, and generating variance reports
Create comprehensive business plans, financial projections, and strategic documents for funding or planning
Calculate cash runway, burn rate, and financial sustainability projections
TradingAgents: Multi-Agents LLM Financial Trading Framework
Build your autonomous hedge fund in minutes. AutoHedge harnesses the power of swarm intelligence and AI agents to automate market analysis, risk management, and trade execution.
💰 Double-entry bookkeeping made easy — plain-text accounting for humans and AI agents. Polished iOS & Android app built with React Native + Expo.
Pizzly, financial market analysis combining technical indicators with LLMs, featuring real-time data processing and AI-powered market insights ⚡️
Learn how to use the Financial Analyst Claude skill. Complete guide with installation instructions and examples.
Learn how to use the helium-mcp Claude skill. Complete guide with installation instructions and examples.
Learn how to use the Runway Calculator Claude skill. Complete guide with installation instructions and examples.