GrowwReview
Ashishbhardwaj97/GrowwReview
Automated weekly review analysis for Groww. Scrapes Play Store reviews, clusters them using ML embeddings, generates AI summaries via Groq, and delivers insights to Google Docs & Gmail via a remote MCP server.
Instalação
npx convez-ai Ashishbhardwaj97/GrowwReview1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
README
A tradução para o português deste README ainda está na fila.
Groww Weekly Review Pulse
Automated system that ingests Groww's Google Play Store reviews, clusters them into semantic themes using ML, and delivers a weekly insight report via Google Docs and Gmail — powered by MCP servers.
Overview
This project automates the weekly review analysis workflow for the Groww mobile app:
- Ingest — Scrape recent reviews from the Google Play Store
- Preprocess — PII scrubbing, deduplication, normalisation
- Embed — Generate semantic embeddings (sentence-transformers)
- Cluster — UMAP dimensionality reduction → HDBSCAN density clustering
- Summarise — LLM-powered theme naming, quote extraction, and action ideas (via Groq)
- Deliver — Append a weekly section to a running Google Doc and email a teaser to stakeholders
Quick Start
# 1. Clone the repository
git clone <repo-url> && cd GrowwReview
# 2. Create a virtual environment
python -m venv .venv && .venv\Scripts\activate # Windows
# python -m venv .venv && source .venv/bin/activate # macOS/Linux
# 3. Install Python dependencies
pip install -r requirements.txt
# 4. Configure environment variables
copy .env.example .env
# Edit .env and add your GROQ_API_KEY
# 5. Review and customise config.yaml
# Update google_doc_id, stakeholder emails, etc.
# 6. Run the pipeline (dry run — no delivery side effects)
python src/cli.py --product groww --week auto --dry-run
Setup Guide & Configuration Reference
Environment Variables (.env)
See .env.example for the required keys.
GROQ_API_KEY: Required for LLM summarisation.MCP_SERVER_URL: Optional. URL to your Google MCP server (overridesconfig.yaml).
config.yaml Reference
delivery.google_doc_id: The ID of the target Google Doc where reports will be appended.delivery.stakeholders: List of email addresses to receive the draft email.mcp_server.url: The Server-Sent Events (SSE) endpoint of the MCP server (e.g.,https://web-production-131a3.up.railway.app/sse).delivery.draft_only: Iftrue, the email will only be created as a draft.falsesends the email.
Remote MCP Server
This system relies on a remote Google MCP Server to perform actions in Google Docs and Gmail securely.
- The server should expose an SSE endpoint (e.g.
/sse) and accept JSON-RPC tool calls. - The current production endpoint is
https://web-production-131a3.up.railway.app/sse. Ensure this server is running and authenticated.
CLI Usage
# Full dry-run (does not connect to MCP, just logs what would happen)
python src/cli.py --product groww --week auto --dry-run
# Draft-only mode (appends to Docs, creates Gmail draft but does not send)
python src/cli.py --product groww --week auto --draft-only
# Force a re-run (ignores the run log and processes again)
python src/cli.py --product groww --week auto --force
Scheduling
This pipeline is designed to run periodically (e.g., weekly on Mondays at 09:00 IST).
A GitHub Actions workflow .github/workflows/weekly-pulse.yml is provided for automatic execution.
To use the GitHub Actions scheduler:
- Go to your GitHub repository settings -> Secrets and variables -> Actions.
- Add a new repository secret named
GROQ_API_KEY. - Optionally add
MCP_SERVER_URLif you want to override the default URL inconfig.yaml. - The workflow will automatically trigger based on the cron schedule (
30 3 * * 1which is Monday 09:00 IST / 03:30 UTC).
Project Structure
See docs/architecture.md for the full architecture and directory layout.