eeye

eeye is a stock screener that leverages a multi-stage pipeline to fetch NSE stocks, pull real-time data from Groww, and apply multiple technical analysis strategies (such as EMA, RSI, Bollinger Bands, and volume-based indicators) to identify potential trading opportunities. When run in MCP server mode, it exposes a standardized interface that AI assistants can query to obtain technical data, OHLC histories, and strategy-driven results. To use the MCP server, start the server with the --mcp flag, which will launch an HTTP API on the port defined in your .env (default localhost:3000). The pipeline processes stocks in parallel: it discovers and backfills data, loads it into an in-memory cache, runs several strategies concurrently, aggregates results, and serves on-demand queries via the MCP interface. You can also run the tool as a standalone stock screener (without MCP) using the default run path to perform screening with optional --cleanup to remove de-listed stocks.

Prerequisites:

• Go 1.20+ (the project notes indicate Go 1.24.5 in this README)
• Docker (for TimescaleDB and database setup, if you prefer containerized DB)
• Groww API access token (GROWW_ACCESS_TOKEN) and optional Groww API base URL

Installation steps:

1. Clone the repository: git clone <repo_url> cd <repo_directory>

2. Install dependencies and environment setup scripts (as indicated by the project):

   • Make scripts executable and run them as described in the README chmod +x scripts/setup-env.sh ./scripts/setup-env.sh

3. Configure Groww and database credentials:

   • Create/modify .env to include GROWW_ACCESS_TOKEN and any other default config values
   • Ensure database connection details (DB_HOST, DB_PORT, DB_NAME, DB_USER, DB_PASSWORD) are set

4. Setup the database (TimescaleDB) via the provided script: chmod +x scripts/setup-db.sh ./scripts/setup-db.sh

5. Build/run the application in MCP mode (as shown in the README):

   • To run as MCP server: go run main.go --mcp

6. Optional: set up pre-commit hooks and development tools as described: chmod +x scripts/install-hooks.sh ./scripts/install-hooks.sh

Notes and tips:

• The MCP server defaults to host:port specified in .env (typically localhost:3000). Ensure your environment exposes this port if you need external access.
• The Groww token must be kept secure; avoid committing tokens to version control.
• When running as a screener (non-MCP), the pipeline will perform data discovery, cache loading, and parallel strategy execution to produce a set of candidate stocks per strategy.
• If you encounter database readiness issues, re-run the database setup script; it’s idempotent and will not duplicate data.
• The system supports a --cleanup flag to remove de-listed stocks after analysis; use it to keep the database lean.
• For MCP integration, ensure the environment variable PORT is aligned with your reverse proxy or client expectations, and that the .env file contains the correct host/port configuration.