Skip to main content

Configure the server

After installing bunny, configure the agent once per machine or container. This page covers bunny configure, config files, network access, and production setup.

Overview​

Install → bunny configure → bunny run → connect (SSH tunnel or public URL)
              │
              ├─ Owner account + MFA
              ├─ ~/.config/bunny/config.yaml
              ├─ SQLite auth DB (~/.config/bunny/)
              └─ Optional: Discord (bunny discord setup)

bunny configure (first run)​

Run once on a fresh install:

bunny configure

Inside Docker:

docker compose exec -it bunny bunny configure

The wizard:

  1. Owner account — email + password (Argon2id). Only one owner at bootstrap; more users via invitations in the Web UI.
  2. MFA (recommended) — TOTP (Google Authenticator, etc.). Recovery codes are shown once — save them.
  3. Public URL — how browsers reach this agent (watch links, OAuth, team chat deep links). Prompted for remote/production deployments.
  4. Discord (optional) — if you accept, runs the same flow as bunny discord setup (Application ID, bot token, OAuth). See Discord setup.

Re-run bunny configure later to change Discord settings or re-bootstrap Discord when tokens rotate.

Non-interactive flags:

bunny configure --email you@example.com --password '…' --public-url https://your-host.example.com

Config files​

FilePurpose
~/.config/bunny/config.yamlMain agent config (server, security, team chats, agents, Discord, terminals)
~/.config/bunny/*.dbAuth, sessions, audit (SQLite)
~/.config/bunny/secrets.encEncrypted secrets vault (optional)
.discord/bridge.yamlDiscord bridge bot token + guild (dev; or path from bunny discord setup)
.bunny.yamlOptional override in repo cwd (dev)

Create a default config.yaml without the full wizard:

bunny config-init

Key settings (config.yaml)​

server:
  bind_host: "127.0.0.1"   # use 0.0.0.0 in Docker
  port: 7681
  data_dir: "~/.config/bunny"
  public_url: "https://your-host.example.com"   # external URL (proxy, HTTPS); distinct from bind_host

terminal:
  shell: "/bin/bash"
  backend: "tmux"          # persistent shells across restarts

browser:
  enabled: true

team_chats:
  link_code_ttl_minutes: 15   # session ↔ channel link codes (all team chat connectors)

agents:
  max_turns: 30               # max turns per agent invocation (Claude today; future agents)

discord:
  enabled: true

team_chats and agents apply to all team chat connectors (Discord today; Slack/Teams planned). Discord-specific credentials (bridge_token_hash, OAuth) stay under discord:.

server.public_url is the URL browsers use to reach the agent (watch links, OAuth callbacks, future team chat deep links). It is often different from bind_host/port when behind a reverse proxy or HTTPS terminator.

Environment variables​

Any YAML key can be overridden with BUNNY_ + nested keys in SCREAMING_SNAKE:

VariableEffect
BUNNY_SERVER__BIND_HOST=0.0.0.0Listen on all interfaces (Docker)
BUNNY_SERVER__PORT=7681HTTP port
BUNNY_SERVER__PUBLIC_URL=https://…Agent public URL (set during bunny configure)
BUNNY_TEAM_CHATS__LINK_CODE_TTL_MINUTES=15TTL for session ↔ channel link codes
BUNNY_AGENTS__MAX_TURNS=30Max agent turns per invocation
BUNNY_SECRETS_PASSPHRASE=…Unlock secrets vault on start

Double underscore __ = nested YAML key.

Start the agent​

bunny run
FlagPurpose
--host 0.0.0.0Required inside Docker for port mapping
--port 7681HTTP port (default 7681)
--no-web-uiAPI only
--no-discord-bridgeDo not auto-start Discord bridge

bunny start is similar but does not build the Web UI — use when dist is already present.

Network access​

Agent binds to localhost on the server:

bunny run --host 127.0.0.1 --port 7681

From your laptop:

ssh -L 7681:127.0.0.1:7681 user@your-server

Open http://127.0.0.1:7681 locally.

Public IP​

bunny run --host 0.0.0.0 --port 7681

Open firewall port 7681, use MFA, prefer HTTPS reverse proxy in production.

Production: systemd​

After native Linux install:

sudo cp infra/systemd/bunny-agent.service /etc/systemd/system/
sudo systemctl enable --now bunny-agent

Expects bunny at /usr/local/bin/bunny. Data and config stay in the service user's ~/.config/bunny/.

Verify​

bunny doctor    # Chromium, Node, sidecars, tmux, web UI
bunny status    # Running terminals, bind address
bunny auth-status

Web UI after configure​

  1. Open the UI (tunnel or public URL).
  2. Log in with the owner account (+ MFA if enabled).
  3. Create a session (project workspace).
  4. Open terminals, preview, or browser tabs.
  5. Invite teammates (session members) from the session UI.

Optional: secrets vault​

bunny secrets init
bunny secrets set OPENAI_API_KEY --scope system
export BUNNY_SECRETS_PASSPHRASE='your-passphrase'

See Security.

Discord next steps​

  1. Discord application and server setup
  2. Discord workflows — linking channels, threads, Claude
  3. Discord slash commands — full /bunny reference

See also​