Introduction
The Idea
Section titled “The Idea”You’re deep in focus mode, but Slack keeps pinging. Some messages need your attention, others don’t. What if you had an AI assistant that could handle the routine stuff — in your voice, following your rules?
ForkMe is a desktop app that acts as your Slack autopilot. It monitors your workspaces and responds to messages automatically using AI, while you stay focused on what matters.
How It Works
Section titled “How It Works”- Connect Slack — Sign in to your workspace(s) via the built-in browser
- Set up AI — Configure one or more AI providers (cloud or local)
- Create Personas — Define how the AI should respond (tone, style, boundaries)
- Define Rules — Specify when and where to auto-respond
- Let it run — ForkMe handles messages while you work
Key Concepts
Section titled “Key Concepts”🤖 AI Providers
Section titled “🤖 AI Providers”ForkMe is provider-agnostic. Use whatever AI works for you:
| Provider | Description |
|---|---|
| AWS Bedrock | Claude models via your AWS account |
| OpenAI | GPT-4, GPT-4o with your API key |
| Anthropic | Direct Claude API access |
| OpenAI-Compatible | Local models via Ollama, LM Studio, or any compatible endpoint |
You can configure multiple providers and switch between them per persona.
🎭 Personas
Section titled “🎭 Personas”Personas define the AI’s personality and behavior. Each persona has:
- System Prompt — Instructions that shape responses (tone, boundaries, knowledge)
- AI Provider & Model — Which model to use
- MCP Servers (optional) — Connect to external tools (Linear, GitHub, etc.)
- Triage Settings — When to escalate instead of auto-respond
Example personas:
- “Work Mode” — Professional, concise, redirects to focus hours
- “Support Agent” — Helpful, thorough, uses documentation tools
- “After Hours” — Brief, polite, defers to next business day
📋 Rules
Section titled “📋 Rules”Rules determine when ForkMe responds. Configure:
- Channels/DMs — Which conversations to monitor
- Users — Only respond to specific people
- Keywords — Trigger on certain words
- @mentions — Only respond when mentioned
- Schedule — Active hours (e.g., weekdays 9-5)
- Priority — Higher priority rules win when multiple match
🚨 Triage
Section titled “🚨 Triage”Not every message should get an auto-response. Triage lets you:
- Automatically detect messages that need human attention
- Set confidence thresholds
- Create escalations for review
- Never miss something important
Privacy & Security
Section titled “Privacy & Security”ForkMe runs entirely on your machine:
- No data sent to ForkMe servers (there are none)
- Slack credentials stay local
- API keys encrypted with OS-level security (Keychain, DPAPI, libsecret)
- You control which AI providers see your messages
Tech Stack
Section titled “Tech Stack”Built with modern tools for reliability and extensibility:
- Electron — Cross-platform desktop shell
- React + TypeScript — Type-safe UI
- Vercel AI SDK — Unified interface for multiple AI providers
- Better-SQLite3 — Fast local storage
- MCP Protocol — Extensible tool integrations
Coming Soon
Section titled “Coming Soon”- Installation guide
- Configuration reference
- Advanced personas tutorial
- MCP integrations guide