AI-powered Slack support assistant with RAG capabilities and a Pi Durable agent runtime. Conversations and admitted requests survive runtime restarts.
- Durable agents: Persisted conversations, crash recovery, and queued Slack replies
- RAG-powered Answers: Uses indexed documents via LangChain + ChromaDB
- Multi-channel Support: DMs, mentions, slash commands
- Document Indexing: Auto-index files shared in Slack
git clone https://github.com/echohello-dev/wingman.git
cd wingman
cp .env.example .env
# Edit .env with your Slack tokens and API keys
docker compose up -dAccess: Backend • Dashboard • API Docs
In Slack:
- DM: Send a message to @Wingman
- Mention:
@Wingman what's up? - Command:
/wingman hello
Wingman posts the final answer in Slack after processing the request.
| Guide | Purpose |
|---|---|
| Getting Started | Quick start |
| Local Development | Run locally with streaming |
| Setup Guide | Full deployment |
| Slack Auth | Token reference |
| Pi Runtime | Persistence, recovery, and local development |
FastAPI backend with Slack bot, Pi Durable agent service, RAG retrieval (LangChain + ChromaDB), PostgreSQL storage, and Next.js dashboard. See ADR-0004 for the runtime decision.
See contributing.md for development workflow, commands, and guidelines.
For AI coding agents: AGENTS.md contains architecture and command reference for automated development.
See ./docs for setup guides and ./docs/adrs for architecture decisions.
Contributions welcome! See contributing.md for guidelines.
See LICENSE for details.
FastAPI • Slack Bolt • LangChain • OpenRouter • ChromaDB • Next.js
Made with care for better Slack support