A personal AI companion you host yourself.
It remembers you, shows its work, and never acts without your say.
Overview · How it works · Using it · Running it · Status
- Talks and remembers. Memories carry their source, and you can see why it knows something.
- Shows its work. Every answer shows which memories, tools and model it used.
- Asks before acting. Actions go through one controlled gate. You approve the exact action, once.
- Stays yours. Runs on your hardware, on your private network, with local models by default.
Status: early. Chat, memory, approvals and workflows work end to end on a real server. Calls, agents and the proposal engine are still preview or planned. Full status
The real interface with demo data, no server needed. Needs Node.js 22.12 or newer:
git clone https://github.com/Conker-AI/conker.git
cd conker/dashboard
npm ci
npm run dev -- --host localhost --port 5173 --strictPortOpen http://localhost:5173 and start chatting. Answers, memories and approvals are simulated in the browser; ask it to email someone to see an approval. To run the real thing on a server, see 4 · Running Conker.
flowchart LR
You([You]) --> GW[Gateway] --> Pi[Pi<br/>conversations]
Pi --> MG[MemoryGate<br/>remembers]
Pi --> TG[ToolGate<br/>acts, with approval]
Pi --> SG[SystemGate<br/>observes]
Pi --> M[Models<br/>local or hosted]
Six small services, each in its own repository and usable on its own: pi · memorygate · toolgate · systemgate · embeddings. More in 2 · How it works.
| 1 · Overview | What Conker is, why, and the 10 words it uses |
| 2 · How it works | Diagrams: services, one message, one approval |
| 3 · Using Conker | Each screen and common tasks |
| 4 · Running Conker | Preview, server, local Windows stack |
| 5 · Developing | Repos, checks, rules, decisions |
| Status | What works today, and what doesn't |
The frontend builds on shadcnstore's dashboard template (notice), React, Vite, Tailwind CSS and Radix.
MIT. Found a vulnerability? Please report it privately: security policy.