Documentation
Nakama is a multi-tenant platform for teams of AI agents. Each profile is a bot with its own role, soul, tools, and memory — exposed through the web dashboard, CLI, Telegram, WhatsApp, or Discord from one deployment.
Quick start
Pick the fastest path for you:
# Live demo — no install
# Open https://demo.getnakama.cloud
# Username: demo@getnakama.cloud
# Password: demo1234
# Managed hosting — your own instance (recommended)
# Sign up at https://getnakama.cloud/ and open your instance URL
# Docker — single container
docker pull ghcr.io/ahmadrosid/nakama:latest
docker run -d -p 4310:4310 -v nakama-data:/nakama/data --name nakama ghcr.io/ahmadrosid/nakama:latest
# Local development — from source
git clone https://github.com/ahmadrosid/nakama.git && cd nakama
bun install && bun run dev:webThen open the dashboard, complete the setup wizard, and send your first message. See Quickstart for the full first-run flow.
Start here
- Quickstart — try the demo, install, and send your first message
- Overview — organizations, profiles, tools, and channels
- First-time setup — admin account, org, provider, and profiles
- Providers — LLM API keys and model configuration
Deploy
- Quickstart — demo, managed hosting, Docker, or local Bun
- Docker — production-style single-container deployment
- Coolify — one-click self-hosting with automatic reverse proxy and SSL
- Deployment hardening — TLS, network exposure, secrets, backups, and verification
- Outbound connections — external destinations, triggers, and data flows
- Live demo — shared instance (
demo@getnakama.cloud/demo1234) - Managed hosting — sign up and open your dedicated URL
- Backup and restore — export and restore your Nakama data root
- Error tracking: ship this instance's own crashes to your Sentry, GlitchTip, Bugsink, or Rustrak project
Concepts
- Multi-tenancy — organizations, members, roles, and tenant isolation
- Org memory — shared, admin-curated facts for the whole organization
- Self-improving skills — agents save workflows as skills; optional admin approval
- Profiles — soul files, memory, tools, and model selection
- Browse Files and artifacts: browse, preview, edit, and publish what your agent saves
- Automations and tasks: scheduled prompts, run history, and channel delivery
- Workflows: on-demand step recipes with receipts and summarize-from-data runs
- Agent prompts — how Nakama builds the system prompt each turn
Channels
- CLI — terminal chat and slash commands
- Telegram — bot setup, pairing, groups, and troubleshooting
- WhatsApp — direct chat linking, groups, commands, and troubleshooting
- Discord — bot setup, pairing, slash commands, and servers
- Slack: app manifest, pairing, DMs, and channel threads
Extend
- Builtin tools — file access, web search, email, bash, and more
- Skills — reusable workflows including memory, artifacts, and automations
- Self-improving skills — agents save workflows as skills; optional admin approval
- Integrations: notifications, Composio, coding agents, context savings, and error tracking
- Image generation: generate images with OpenAI gpt-image-2
- Token optimiser: shorten long tool output and measure what it saves
- Coding agent — Codex, Claude Code, OpenCode, pi, or Cursor Agent from chat
- Agent browser — interactive browser automation for login-walled sites
- MCP servers — connect external tool providers
- Composio — SaaS OAuth and toolkit assignment
Community
- Contributing — run from source, test, and open a pull request
- Changelog: what changed in each release, and how to read it
- GitHub repository — source code and issues
Reference
- llms.txt — documentation index for AI agents and tooling
- Threat model — STRIDE review by architecture component
- GitHub repository — source code and issues