Commands Reference

Complete reference for all Fulcrum slash commands

Fulcrum provides slash commands that orchestrate your AI team. These commands are the primary interface for directing work, managing tasks, and coordinating agents.

💡 Pro tip: You don't need to memorize all commands. Start with /start for any new work—it will guide you to the right workflow.

Session Commands

Commands for starting work and checking system status.

Command	Description	Parameters
`/start`	Universal entry point for any work. Routes to appropriate workflow based on work type.	Work description (required)
`/status`	Verify Fulcrum configuration and system health. Shows active agents and setup state.	None

/start

The /start command is your universal entry point. It analyzes your request and routes to the appropriate workflow:

Greenfield projects → Inception workflow with PM & Tech Lead
New features → PRD creation → Technical design → Execution
Technical improvements → Direct to Tech Lead for specification
Bug fixes → Triage → Investigation → Fix

# Start a new feature
/start "Add user authentication with OAuth"

# Start a greenfield project
/start "Build a SaaS analytics platform"

# Start a technical improvement
/start "Refactor the payment module for better testability"

📝 Best practice: Be specific about what you want to achieve. The more context you provide, the better the orchestrator can route your request.

Work Item Commands

Commands for creating and managing work items in the Beads issue tracker.

Command	Description	Parameters
`/create-work-item`	Create a new trackable work item in Beads.	Title, type, priority, description
`/start-work-item`	Begin working on an assigned work item.	Work item ID
`/update-work-item`	Update progress, status, or details of a work item.	Work item ID, fields to update
`/close-work-item`	Mark work as complete. Requires peer review.	Work item ID, reason
`/handoff-work-item`	Transfer ownership between agents with context.	Work item ID, target agent

/create-work-item

Creates a new work item with full tracking:

# Create a feature work item
/create-work-item "Add OAuth integration" type:feature priority:P1

# Create with dependencies
/create-work-item "Build login UI" type:feature parent:bd-auth.1

# Create a bug fix
/create-work-item "Fix token refresh" type:bug priority:P0

Work item types:

feature — New functionality
bug — Defect fix
task — General task
chore — Maintenance work
spike — Research/investigation

/start-work-item

Begins execution on an assigned work item:

# Start working on a specific item
/start-work-item bd-auth.3

# The agent will:
# 1. Load all context from the work item
# 2. Check dependencies are satisfied
# 3. Begin implementation

/update-work-item

Updates work item state and progress:

# Update status
/update-work-item bd-auth.3 status:in_progress

# Update priority
/update-work-item bd-auth.3 priority:P0

# Add blockers
/update-work-item bd-auth.3 blocked_by:bd-auth.2

/close-work-item

# Close with reason (required)
/close-work-item bd-auth.3 --reason "Implemented and all tests passing"

# Note: Peer review is required before closing

⚠️ Important: Work items require peer review before closing. Use /request-review first if you haven't already.

Execution Commands

Commands for planning and executing work across multiple agents.

Command	Description	Parameters
`/plan-execution`	Analyze requirements and create executable work breakdown.	`project:[name]`
`/execute-work`	Dispatch work items to agents in parallel based on dependencies.	`project:[name]`, `all`, `resume`

/plan-execution

Takes a PRD or technical design and creates an executable plan with Beads work items:

# Plan from a project specification
/plan-execution project:user-authentication

# The Tech Lead will:
# 1. Analyze the PRD against your codebase
# 2. Identify technical gaps and risks
# 3. Create Beads work items with dependencies
# 4. Assign appropriate agents to each item

/execute-work

The powerhouse command. Dispatches ready work items to agents in parallel:

# Execute all ready items in a project
/execute-work project:auth

# Execute all ready items across all projects
/execute-work all

# Resume after interruption
/execute-work resume

# Execute multiple projects
/execute-work projects:auth,payments,notifications

💡 How parallel execution works: The orchestrator identifies items with satisfied dependencies, dispatches them to available agents, and monitors progress. As items complete, newly unblocked items are automatically picked up.

Review Commands

Commands for code review and quality assurance.

Command	Description	Parameters
`/request-review`	Initiate peer review council before QA handoff.	Work item ID, options

/request-review

Initiates the peer review process with the council pattern:

# Request review for a work item
/request-review bd-auth.3

# Skip review for documentation-only changes
/request-review bd-docs.1 --skip-review --reason "docs-only"

# Force review even if criteria suggests skip
/request-review bd-auth.3 --force-review

The review council includes:

Peer reviewer for code quality
Security review for sensitive changes
Tech Lead for architectural decisions

Setup Commands

Commands for initial setup and configuration.

Command	Description	Parameters
`/setup-fulcrum`	Intelligent setup with repo analysis and agent optimization.	None (interactive)
`/setup-beads`	Initialize Beads distributed issue tracker.	`--with-agent-mail`, `--force`, `--status`
`/sync-handbook`	Sync project documentation to organization-wide handbook.	None
`/generate-agent`	AI-powered creation of specialized agents.	None (interactive)
`/manage-agents`	Add or remove agents from your team.	None (interactive menu)

/setup-fulcrum

Run this once after installation. It analyzes your repository and configures optimal settings:

# Run intelligent setup
/setup-fulcrum

# The setup will:
# 1. Analyze your codebase structure
# 2. Detect languages and frameworks
# 3. Recommend appropriate agents
# 4. Configure vendor integrations
# 5. Initialize Memory Bank

/setup-beads

Initializes the Beads distributed issue tracker for work item management:

# Basic setup
/setup-beads

# With agent mail for async notifications
/setup-beads --with-agent-mail

# Force re-initialization
/setup-beads --force

# Check current status
/setup-beads --status

/generate-agent

Creates custom specialized agents for your specific needs:

# Start interactive agent generation
/generate-agent

# Agent types available:
# - Module specialist (e.g., auth-specialist)
# - Technology expert (e.g., graphql-expert)  
# - Discipline agent (e.g., accessibility-engineer)
# - QA specialist (e.g., security-tester)
# - Custom (define your own)

Quick Reference

I want to...	Use this command
Start any new work	`/start "description"`
Execute planned work	`/execute-work project:name`
Create a work item manually	`/create-work-item "title" type:feature`
Check system health	`/status`
Get code reviewed	`/request-review bd-xxx`
Add a custom agent	`/generate-agent`
Initial setup	`/setup-fulcrum`

Workflow Patterns

Feature Development

/start "Add user authentication with OAuth"
# → PM creates PRD
# → Tech Lead creates design + Beads items
/execute-work project:auth
# → Agents implement in parallel
/request-review bd-auth.3
# → Peer review
/close-work-item bd-auth.3 --reason "Complete"

Bug Fix

/start "Fix: Users can't reset passwords"
# → Creates investigation task
/start-work-item bd-bug.1
# → Agent investigates and fixes
/request-review bd-bug.1
/close-work-item bd-bug.1 --reason "Fixed"

Greenfield Project

/start "Build a real-time collaboration tool"
# → Inception workflow
# → PM + Tech Lead discovery
# → Foundation setup
/plan-execution project:collab-tool
/execute-work project:collab-tool

📝 Note: Commands integrate with the Beads CLI under the hood. You can also use bd commands directly for advanced workflows.