Skip to main content

​
Commands

The Compound Engineering plugin provides 22 slash commands organized around a core workflow: brainstorm → plan → work → review → compound.

​
What Are Commands?

Commands are slash commands you invoke in your conversation with Claude. They orchestrate complex workflows by spawning agents, running skills, and coordinating multi-step processes. Invocation: 
/ce:plan Add user authentication
/ce:work docs/plans/2026-03-03-feat-auth-plan.md
/ce:review

​
Core Workflow Commands

These commands use the ce: prefix to unambiguously identify them as compound-engineering commands. This distinguishes them from Claude’s built-in /plan and /review commands.
Why ce:? The namespace (short for compound-engineering) makes it immediately clear these commands belong to this plugin and provides a distinct workflow separate from Claude’s defaults.
CommandDescriptionUsage Example
/ce:brainstormExplore requirements and approaches before planning/ce:brainstorm user authentication system
/ce:planCreate implementation plans/ce:plan Fix checkout race condition
/ce:workExecute work items systematically/ce:work docs/plans/2026-03-03-feat-auth-plan.md
/ce:reviewRun comprehensive code reviews/ce:review
/ce:compoundDocument solved problems to compound team knowledge/ce:compound

​
Workflow Progression

  1. /ce:brainstorm - Answer WHAT to build through collaborative dialogue
  2. /ce:plan - Answer HOW to build it with detailed planning
  3. /ce:work - Execute the plan with systematic implementation
  4. /ce:review - Review the implementation with multiple specialized agents
  5. /ce:compound - Document the solution for future reference
# Step 1: Explore the idea
/ce:brainstorm API rate limiting for our SaaS product

# Claude asks questions, you collaborate on approach
# Brainstorm saved to docs/brainstorms/2026-03-03-api-rate-limiting-brainstorm.md

# Step 2: Create detailed plan
/ce:plan  # Auto-detects recent brainstorm

# Research agents gather context, plan saved to docs/plans/2026-03-03-feat-api-rate-limiting-plan.md

# Step 3: Implement the plan
/ce:work docs/plans/2026-03-03-feat-api-rate-limiting-plan.md

# Code is written, tests are created, commits are made

# Step 4: Review the implementation
/ce:review

# Security, performance, Rails reviewers all provide feedback
# Address feedback, iterate

# Step 5: Document the solution
/ce:compound

# Solution saved to docs/solutions/performance-issues/api-rate-limiting.md
# Available for future reference via learnings-researcher agent

​
Deprecated Aliases

The following aliases still work but show deprecation warnings:
  • /plugin/workflow:plan → Use /ce:plan
  • /plugin/workflow:work → Use /ce:work
  • /plugin/workflow:review → Use /ce:review
  • /plugin/workflow:brainstorm → Use /ce:brainstorm
  • /plugin/workflow:compound → Use /ce:compound
Update to the ce:* equivalents to avoid warnings.

​
Utility Commands

Utility commands automate specific tasks and support the core workflow.

​
Autonomous Workflows

CommandDescriptionUsage
/lfgFull autonomous engineering workflow/lfg Add OAuth2 support
/slfgFull autonomous workflow with swarm mode for parallel execution/slfg Refactor authentication system
Both run the complete workflow: plan → deepen → work → review → test → video. /slfg uses swarm orchestration for parallel agent execution.
/lfg (“Let’s Fucking Go”) runs this sequence automatically:
  1. /ce:plan - Create implementation plan
  2. /deepen-plan - Enhance plan with research agents
  3. /ce:work - Execute the plan
  4. /ce:review - Run code reviews
  5. /resolve_todo_parallel - Address review feedback
  6. /test-browser - Test affected pages
  7. /feature-video - Record demo video
  8. Outputs completion signal when done
Use when you want full automation from idea to PR with video demo.

​
Planning Enhancements

CommandDescriptionUsage
/deepen-planEnhance plans with parallel research agents for each section/deepen-plan docs/plans/2026-03-03-feat-auth-plan.md
What it does:
  • Spawns research agents for every plan section
  • Discovers and applies ALL available skills
  • Checks institutional learnings (docs/solutions/)
  • Queries Context7 for framework documentation
  • Runs ALL review agents in parallel
  • Synthesizes findings into enhanced plan
See deepen-plan details for more.

​
Creation & Documentation

CommandDescriptionUsage
/changelogCreate engaging changelogs for recent merges/changelog weekly
/create-agent-skillCreate or edit Claude Code skills/create-agent-skill
/generate_commandGenerate new slash commands/generate_command analyze-dependencies
/heal-skillFix skill documentation issues/heal-skill skill-name
# 🚀 Weekly Change Log: March 3, 2026

## 🌟 New Features
- Added OAuth2 authentication support (#456)
- Implemented API rate limiting with Redis (#457)

## 🐛 Bug Fixes
- Fixed race condition in checkout flow (#458)
- Resolved N+1 queries in briefs index (#459)

## 🙌 Shoutouts
@alice for the OAuth2 implementation
@bob for performance improvements

## 🎉 Fun Fact of the Day
We reduced API response time by 60% this week!

​
Bug Management

CommandDescriptionUsage
/report-bugReport a bug in the plugin/report-bug /ce:work fails with empty plan
/reproduce-bugReproduce bugs using logs and console/reproduce-bug #123
/triageTriage and prioritize issues/triage

​
Parallel Resolution

CommandDescriptionUsage
/resolve_parallelResolve TODO comments in parallel/resolve_parallel
/resolve_pr_parallelResolve PR comments in parallel/resolve_pr_parallel
/resolve_todo_parallelResolve todos in parallel/resolve_todo_parallel
These commands spawn multiple agents to address feedback simultaneously, dramatically speeding up iteration.

​
Testing & Validation

CommandDescriptionUsage
/test-browserRun browser tests on PR-affected pages/test-browser
/xcode-testBuild and test iOS apps on simulator/xcode-test
/feature-videoRecord video walkthroughs and add to PR description/feature-video
/test-browser requires the agent-browser CLI tool:
npm install -g agent-browser
agent-browser install  # Downloads Chromium
See the agent-browser skill for details.

​
Other Utilities

CommandDescriptionUsage
/agent-native-auditAudit features for agent-native design compliance/agent-native-audit
/deploy-docsDeploy documentation sites/deploy-docs

​
Command Invocation Patterns

​
With Arguments

Most commands accept arguments to specify what to work on: 
/ce:plan Add user authentication
/ce:brainstorm API rate limiting strategy
/changelog daily

​
With File Paths

Some commands work with files: 
/ce:work docs/plans/2026-03-03-feat-auth-plan.md
/deepen-plan docs/plans/2026-03-03-feat-auth-plan.md

​
Without Arguments

Commands without arguments often prompt for input or auto-detect context: 
/ce:review              # Reviews current branch changes
/ce:compound            # Prompts for problem description
/test-browser           # Detects PR-affected pages

​
Background Execution (Claude Code Web)

Append & to run commands in the background on Claude Code web: 
/ce:work docs/plans/2026-03-03-feat-auth-plan.md &
This lets you continue other work while the command runs.

​
Command Configuration

Some commands respect configuration in CLAUDE.md:

​
Review Configuration

Configure which review agents run: 
# In CLAUDE.md
review_agents:
  - security-sentinel
  - performance-oracle
  - kieran-rails-reviewer
See the setup skill for interactive configuration.

​
Project Tracker

Configure issue creation from plans: 
# In CLAUDE.md
project_tracker: github  # or 'linear'
Used by /ce:plan when creating issues.

​
Command Locations

Commands are organized in the plugin directory: 
plugins/compound-engineering/commands/
├── ce/                      # Core workflow commands
│   ├── brainstorm.md
│   ├── plan.md
│   ├── work.md
│   ├── review.md
│   └── compound.md
├── workflows/               # Deprecated aliases (forward to ce/*)
│   ├── brainstorm.md
│   ├── plan.md
│   ├── work.md
│   ├── review.md
│   └── compound.md
└── *.md                     # Utility commands
    ├── lfg.md
    ├── deepen-plan.md
    ├── changelog.md
    ├── test-browser.md
    └── ...

​
Creating Custom Commands

Use /generate_command to create your own commands: 
/generate_command analyze-dependencies "Analyze and report on project dependencies"
This creates a new command file with boilerplate structure. Commands are markdown files with YAML frontmatter: 
---
name: analyze-dependencies
description: Analyze and report on project dependencies
argument-hint: "[package name]"
---

# Analyze Dependencies

[Command implementation]

​
See Also

  • Agents - Specialized agents that commands orchestrate
  • Skills - Knowledge domains that commands can activate
  • Workflows Guide - End-to-end workflow examples