Slash Commands Reference

Slash commands live in .claude/commands/ and are the prompts that agents execute. They are the system — improving a command improves every future run that uses it.

Core SDLC Commands

Command	Purpose	Called by
/classify_issue	Classify as /chore, /bug, or /feature	adw_plan.py
/feature	Generate feature implementation plan	adw_plan.py (feature)
/bug	Generate bug fix plan with repro steps	adw_plan.py (bug)
/chore	Generate maintenance/refactor plan	adw_plan.py (chore)
/implement	Implement a plan from specs/	adw_build.py
/test	Run unit + integration tests	adw_test.py
/test_e2e	Run E2E tests via Playwright MCP	adw_test.py
/commit	Stage and commit with semantic message	adw_build.py
/pull_request	Create PR with description	adw_plan.py
/review	Review implementation against spec	adw_review.py
/document	Generate technical documentation	adw_document.py
/resolve_failed_test	Debug and fix failing tests	adw_test.py
/resolve_failed_e2e_test	Debug failing E2E tests	adw_test.py

Task System Commands

Command	Purpose
/process_tasks	Analyze tasks.md → return eligible tasks as JSON
/init_worktree	Create sparse-checkout git worktree
/mark_in_progress	Update [] → [in progress, adw_id]
/update_task	Update [in progress] → [done] or [failed]
/build	Implement task directly (no planning phase)
/plan	Create plan in specs/plan-{adw_id}-*.md
/clean_worktree	Remove worktree and its branch

Notion Prototype Commands

Command	Purpose
/get_notion_tasks	Query Notion DB for eligible tasks
/update_notion_task	Update Notion page status
/plan_uv_script	Plan UV single-file Python script MVP
/plan_vite_vue	Plan Vite + Vue 3 web application MVP
/plan_bun_scripts	Plan Bun TypeScript application MVP
/plan_uv_mcp	Plan MCP server for AI tool integration

Running Commands Directly

Use adw_slash_command.py to run any slash command from the terminal:

# Run the /chore command with a task description
uv run adws/adw_slash_command.py /chore "Add error handling to all endpoints"

# Run /implement with a specific spec file
uv run adws/adw_slash_command.py /implement specs/my-plan.md --model opus

# Run /feature with a description
uv run adws/adw_slash_command.py /feature "Add CSV export to results table"

How Slash Commands Work

Each command is a Markdown file in .claude/commands/. When executed:

1. Claude Code reads the command file
2. The $ARGUMENTS placeholder is replaced with your input
3. Claude Code executes as if you typed the prompt manually

Example command file structure (feature.md):

## Your task

Generate a detailed implementation plan for this feature request:

$ARGUMENTS

## Requirements

Include:
- Technical approach
- Step-by-step implementation tasks
- Files to create/modify
- Testing requirements
- Acceptance criteria

## Context

The spec should be specific enough that another agent can implement
it without asking questions.

## Relevant Files
[Update this section to match your project structure]
- src/components/
- src/hooks/
- src/api/

Improving Commands

When an agent produces bad output for a certain type of task, fix the command:

# Edit the command that's producing bad results
$EDITOR .claude/commands/feature.md

# The fix propagates to all future runs

Common improvements:

• Add project-specific file structure to "Relevant Files"
• Add coding conventions ("always use named exports", "hooks go in src/hooks/")
• Add anti-patterns ("never use any from third-party APIs without Zod validation")
• Add specific acceptance criteria templates for your domain