How to discover what BMO can do
BMO exposes its capabilities through several discovery surfaces. The important thing is not to memorize them all. Learn the two or three that match the way you work.
| Mechanism | What it is |
|---|---|
| Onboarding | On first run, BMO can initialize project context with /init, then point you at the core entry surfaces instead of dropping you into a blank screen. |
| Product guide | Start with Overview, How BMO thinks and acts, and Workflow map. |
| UI hints | The landing view and empty-chat state explicitly steer you to /help, /tools, /capabilities, and /discovery; the status bar and command palette extend that guidance without leaving the TUI. Set options.tui.show_landing_hints = false to hide the initial landing hints. |
| Agent self-description | The prompt stack and built-in capability surfaces tell the agent what BMO can do, so asking “what tools do you have here?” is a valid move. |
| Suggested prompts | The landing and empty-chat views expose common starter prompts such as code explanation, edits, and test runs. The landing choices follow options.tui.show_landing_hints. |
| Slash commands | Type / or press ctrl+p for /help, /tools, /capabilities, /discovery, /agent_kinds, /mode, and other operator surfaces. |
| Reference pages | Tool, command, config, and parity docs let you inspect concrete surfaces once you know which part of the system you care about. |
The highest-signal discovery loop
Section titled “The highest-signal discovery loop”- Start with Quickstart.
- Start from the landing or empty-chat cues, then use
/help,/tools,/capabilities, and/discoveryin the TUI. - When you need the mental model, read How BMO thinks and acts and Workflow map.
- When you need exact contracts, move to Tools Reference, Slash Commands, and Configuration.
Pick by maturity
Section titled “Pick by maturity”Use the smallest surface that matches your job:
| If the capability is… | What to do |
|---|---|
| Daily-use | Learn it from guides and feature pages; it belongs in normal operator workflows. |
| Advanced | Use it after the workflow map makes sense, especially for scripts, teams, integrations, long-lived runs, or proposal review before workspace mutation. |
| Maintainer-facing | Treat it as support or contributor material; start from debugger, observability, or source links. |
| Eval-only | Read it as measurement evidence. It does not grant permission, bypass review, or change runtime authority on its own. |
| Reference | Use it for exact commands, config keys, payload fields, and parity contracts after you know the surface you need. |
Related
Section titled “Related”Capability surface map
Section titled “Capability surface map”Capability surfaces
TUIInteractive runs, sidebar state, dialogs, slash commands, debugger.
CLIOne-shot prompts, recipes, schedules, config inspection, diagnostics.
HTTPServer APIs, edit capsules, OpenAI-compatible endpoints, observability.
ProtocolsMCP, A2A, ACP, LSP, and editor integrations.
ToolsFile, shell, LSP, memory, web, workflow, checkpoint, and team actions.
AutomationRecipes, scheduler jobs, rules, autopilot, compaction, hooks.
StateSessions, messages, events, workstreams, checkpoints, artifacts.
DocsQuick guide, operator manual, references, and implementation topics.