Onboard Wizard
The golembot onboard command provides an 8-step interactive wizard for setting up a new assistant from scratch.
Usage
mkdir my-bot && cd my-bot
golembot onboardOr skip interactivity with the --template flag:
golembot onboard --template customer-supportThe 8 Steps
Step 1: Select Engine
Choose your Coding Agent backend:
- cursor — Cursor's
agentCLI - claude-code — Anthropic's
claudeCLI - opencode — OpenCode CLI (multi-provider support)
- codex — OpenAI Codex CLI
Step 2: Engine Authentication
The wizard checks whether the selected engine is already authenticated (via environment variable or CLI login). If not, you can:
- Enter an API key — saved to
.envautomatically - Confirm CLI login — for engines that support browser-based OAuth (
claude auth login,codex login,agent login) - Skip — configure later (the wizard will remind you)
| Engine | API Key | CLI Login |
|---|---|---|
| Cursor | CURSOR_API_KEY | agent login |
| Claude Code | ANTHROPIC_API_KEY | claude auth login |
| OpenCode | Depends on provider | — |
| Codex | CODEX_API_KEY | codex login |
Step 3: Name Your Assistant
Give your assistant a descriptive name (e.g., sales-analyst, team-helper).
Step 4: Select IM Channels
Multi-select which IM platforms to connect (optional). All 6 platforms are supported:
- Feishu (Lark) — WebSocket, no public IP needed
- DingTalk — Stream mode, no public IP needed
- WeCom (WeChat Work) — WebSocket, no public IP needed
- Slack — Socket Mode, no public IP needed
- Telegram — Polling, no public IP needed
- Discord — Gateway API, no public IP needed
Steps 5–6: Configure Channel Credentials
For each selected channel, the wizard prompts for the required credentials:
- Feishu: App ID + App Secret
- DingTalk: Client ID + Client Secret
- WeCom: Corp ID + Agent ID + Secret + Token + Encoding AES Key
- Slack: Bot Token + App-Level Token
- Telegram: Bot Token
- Discord: Bot Token
All credentials are stored in .env with ${ENV_VAR} references in golem.yaml.
Step 7: Choose a Scenario Template
Select from 6 pre-built templates, or choose None to skip:
| Template | Description |
|---|---|
customer-support | FAQ-based support with escalation tracking |
data-analyst | Data analysis with reports and calculations |
code-reviewer | 5-dimension code review with severity tiers |
ops-assistant | Content operations, scheduling, competitor tracking |
meeting-notes | Structured minutes with action item tracking |
research | Structured research reports with source management |
| (None) | Skip — use built-in skills only |
Templates are optional. Selecting "None" skips template installation; the assistant still gets the general and im-adapter built-in skills and is fully functional.
If no template fits your scenario, skip here and create a custom skill later — add a directory under skills/ with a SKILL.md file. See Create a Skill.
Each template includes a tailored golem.yaml, skill directory, and supporting files.
Step 8: Generate & Launch
The wizard generates:
golem.yaml— assistant configuration.env— environment variables (credentials).env.example— template for sharing.gitignore— excludes.golem/and.envskills/— built-in skills + template skillAGENTS.md— auto-generated agent context
Optionally starts the gateway immediately.
Options
| Option | Description |
|---|---|
-d, --dir <dir> | Working directory (default: .) |
--template <name> | Skip template selection step |