Command Overview
Bloom provides commands for workspace setup, repository management, project workflow, task execution, and monitoring.
Command Categories
Bloom has three types of commands:
- Human Commands — You run these directly
- Agent Commands — AI agents use these during task execution
- Internal Commands — Called automatically by other commands or the orchestrator
Most users only need the Human Commands. Agent and Internal commands are documented for completeness.
Human Commands
These are the commands you'll use regularly.
Essential Workflow
| Command | Description |
|---|---|
bloom init | Initialize a new workspace |
bloom repo clone <url> | Clone a repository |
bloom create <name> | Create a new project |
bloom plan | Generate implementation plan |
bloom generate | Generate tasks from plan |
bloom run | Start the multi-agent orchestrator |
Monitoring & Management
| Command | Description |
|---|---|
bloom list [status] | List tasks, optionally filtered |
bloom show <taskid> | Show task details |
bloom dashboard | Live task monitoring |
bloom validate | Validate task configuration |
bloom questions | Show pending questions |
bloom answer <qid> <response> | Answer agent question |
Repository Operations
| Command | Description |
|---|---|
bloom repo list | List all repositories |
bloom repo sync | Sync all configured repos |
bloom repo remove <name> | Remove a repository |
bloom repo worktree add | Add a git worktree |
bloom repo worktree remove | Remove a git worktree |
bloom repo worktree list | List worktrees |
Configuration
| Command | Description |
|---|---|
bloom config | View user configuration |
bloom config set-protocol | Set git protocol preference |
bloom version | Show version |
bloom help | Show help |
Advanced
| Command | Description |
|---|---|
bloom refine | Interactively refine project docs |
bloom enter | Enter Claude Code with project context |
bloom reset <taskid> | Reset stuck task |
bloom reset --stuck | Reset all stuck tasks |
bloom interject | Interrupt running agent sessions |
Agent Commands
These commands are used by AI agents during task execution. You typically don't need to run these manually.
Task Status (Agent Use)
| Command | Description |
|---|---|
bloom done <taskid> | Mark task complete |
bloom block <taskid> | Mark task blocked (needs help) |
bloom note <taskid> <note> | Add note to task |
Questions (Agent Use)
| Command | Description |
|---|---|
bloom ask <agent> <question> | Ask human a question |
bloom wait-answer <qid> | Wait for answer (blocks) |
Internal Commands
These are called automatically by other commands or the orchestrator. Listed for completeness.
Auto-Called by Orchestrator
| Command | Called By | Purpose |
|---|---|---|
bloom ready <taskid> | bloom run | Mark task ready when deps complete |
bloom start <taskid> | bloom run | Mark task when agent begins |
bloom assign <taskid> <agent> | bloom run | Assign to specific agent |
bloom next [agent] | bloom run | Find available tasks |
bloom agents | bloom run | Track agent status |
Auto-Called by Other Commands
| Command | Called By | Purpose |
|---|---|---|
bloom setup | bloom init | Initial repo sync |
bloom repo create | Internal | Create local repos |
Utility (Rarely Needed)
| Command | Purpose |
|---|---|
bloom agent run <name> | Run single agent (debugging) |
bloom agent list | List agent configs |
bloom todo <taskid> | Reset status to todo |
bloom clear-answered | Clean up question queue |
bloom questions-dashboard | Interactive question UI |
Global Options
| Option | Description |
|---|---|
-f, --file <path> | Use custom tasks file |
-l, --log-level <level> | Set log level (debug, info, warn, error) |
Command Details
See individual command pages for detailed usage:
- init — Workspace initialization
- repo — Repository management
- create — Project creation
- refine — Interactive document refinement
- plan — Plan generation
- generate — Task generation
- run — Agent orchestration
- enter — Open-ended Claude sessions
- Task Management — Status and operations
- Questions — Human queue management