Professional-grade workflow automation for AI coding assistants
A cross-platform plugin providing powerful, zero-configuration slash commands for development workflows. Works with Claude Code, Codex CLI, and OpenCode.
📋 Disclaimer: This project originated from personal workflow needs and was made public due to its effective delivery. Usage is entirely at your own responsibility. The maintainers make no guarantees about fitness for any particular purpose. Context/token efficiency has not been formally benchmarked.
💡 Model Recommendation: Using Opus as the main agent model produces significantly better results and follows workflow phases more tightly. While Sonnet works for simpler tasks, Opus is recommended for complex multi-step workflows.
- CLI Installer Fixes - Fixed OpenCode and Codex installation paths
- Codex Skills - Proper
SKILL.mdformat withnameanddescriptionfields - Smaller Package - Removed adapters and dev scripts from npm package
- CLI Installer -
npm install -g awesome-slash@latest && awesome-slashfor cross-platform setup - Reality Check Refactor - Replaced 4 LLM agents with JS collectors (~77% token reduction)
- Automated Releases - GitHub Actions workflow with npm provenance
- Breaking -
.claude/reality-check.local.mdsettings file no longer used
- Platform-Aware State Directories - State now stored in
.opencode/for OpenCode,.codex/for Codex - Fixed OpenCode/Codex Installers - Correct config formats and Windows path handling
- MCP Server Bug Fixes - Fixed workflow state references and resume logic
- Documentation Updates - Added note that Codex uses
$prefix instead of/
/plugin marketplace add avifenesh/awesome-slash
/plugin install next-task@awesome-slash
/plugin install ship@awesome-slashnpm install -g awesome-slash@latest
awesome-slashInteractive installer for Claude Code, OpenCode, and Codex CLI. Select one or more platforms.
Update: npm update -g awesome-slash
Remove: npm uninstall -g awesome-slash
See docs/INSTALLATION.md for all options.
Platform Note: Commands use
/prefix in Claude Code and OpenCode, but$prefix in Codex CLI (e.g.,$next-taskinstead of/next-task).
Complete task-to-production automation with state management and resume capability.
/next-task # Start new workflow with policy selection
/next-task --status # Check current workflow state
/next-task --resume # Resume from last checkpoint
/next-task --abort # Cancel workflow and cleanup
/next-task bug # Filter by task typeWorkflow phases (tracked in .claude/flow.json):
- policy-selection
- task-discovery
- worktree-setup
- exploration
- planning
- user-approval
- implementation
- review-loop
- delivery-approval
- ship-prep
- create-pr
- ci-wait
- comment-fix
- merge
- production-ci
- deploy
- production-release
- complete
Quality gates:
- deslop-work
- test-coverage-checker
- review-orchestrator
- delivery-validator
- docs-updater
Task Sources:
- GitHub Issues - Uses
ghCLI (handles large backlogs with priority filtering) - GitLab Issues - Uses
glabCLI - Local files - Reads from PLAN.md, tasks.md, or TODO.md
- Custom CLI - Any CLI tool (tea, jira-cli, etc.) with auto-discovery
- Other - Describe your source and the agent figures it out
Your source preference is cached in .claude/sources/preference.json for fast subsequent runs.
Notes:
- Fully autonomous after plan approval
- Resume capability with
.claude/flow.json - Policy-based stopping points (pr-created, merged, deployed, production)
- /ship handles PR creation, CI monitoring, merge, and cleanup
Ship your code from commit to production with full validation and state integration.
/ship # Default workflow
/ship --strategy rebase # Rebase before merge
/ship --dry-run # Show plan without executing
/ship --state-file PATH # Integrate with next-task workflowStages:
- Pre-flight checks and platform detection
- Commit and PR creation
- CI wait and review loop
- Merge and (optional) deploy validation
- Cleanup and completion report
Platform Support:
- CI: GitHub Actions, GitLab CI, CircleCI, Jenkins, Travis CI
- Deployment: Railway, Vercel, Netlify, Fly.io, Platform.sh, Render
Remove debugging code, old TODOs, and AI slop from your codebase.
/deslop-around # Report mode - analyze only
/deslop-around apply # Apply fixes with verification
/deslop-around apply src/ 10 # Fix up to 10 issues in src/Detects:
- Console debugging (
console.log,print(),dbg!()) - Old TODOs and commented code
- Placeholder text, magic numbers
- Empty catch blocks, disabled linters
- Placeholder functions (
return 0,todo!(),raise NotImplementedError) - Excessive documentation (JSDoc >3x function body)
- Phantom references (issue/PR mentions, file path references in comments)
- Infrastructure components configured but never used (unused DB clients, caches, API clients)
- Code smells: boolean blindness, message chains, mutable globals, dead code, shotgun surgery
Comprehensive code review with specialized agents that iterate until zero issues.
/project-review # Full codebase review
/project-review --recent # Only recent changes
/project-review --domain security # Domain-focused reviewReview domains: Security, Performance, Architecture, Testing, Error Handling, Code Quality, Type Safety, Documentation
Sync documentation with actual code state across the repository.
/update-docs-around # Report mode - analyze only
/update-docs-around --apply # Apply safe fixes
/update-docs-around docs/ --apply # Sync specific directoryChecks:
- Outdated code references in documentation
- Invalid syntax in code examples
- Missing CHANGELOG entries
- Version mismatches
- Broken file/import paths
Validate task completion and approve for shipping (standalone or part of workflow).
/delivery-approval # Validate current work
/delivery-approval --task-id 142 # Validate specific task
/delivery-approval --verbose # Show detailed check outputValidation checks:
- Tests pass
- Build passes
- Lint passes
- Type check passes
- Task requirements met
Deep repository analysis to identify where documented plans diverge from actual code reality.
/reality-check:scan # Full scan (default)
/reality-check:scan --sources github,docs # Specific sources
/reality-check:scan --depth quick # Quick scanArchitecture:
- JavaScript collectors - Pure JS data collection (no LLM overhead)
- Single Opus call - Deep semantic analysis with full context
- ~77% token reduction - Efficient compared to multi-agent approach
All platforms share the same workflow tools via MCP (Model Context Protocol):
| Tool | Description |
|---|---|
workflow_status |
Get current workflow state |
workflow_start |
Start a new workflow |
workflow_resume |
Resume from checkpoint |
workflow_abort |
Cancel and cleanup |
task_discover |
Find and prioritize tasks |
review_code |
Run pattern-based code review |
See docs/CROSS_PLATFORM.md for details.
Simple state tracking with platform-aware directories:
| Platform | State Directory |
|---|---|
| Claude Code | .claude/ |
| OpenCode | .opencode/ |
| Codex CLI | .codex/ |
Override with AI_STATE_DIR environment variable.
Main project: {state-dir}/tasks.json - Tracks active worktree/task:
{
"active": {
"worktree": "../project-task-123",
"branch": "feature/123-fix-auth",
"taskId": "123",
"taskTitle": "Fix auth timeout"
}
}Worktree: {state-dir}/flow.json - Tracks workflow progress:
{
"task": { "id": "123", "title": "Fix auth timeout" },
"policy": { "stoppingPoint": "merged" },
"phase": "implementation",
"status": "in_progress",
"exploration": { "keyFiles": [...] },
"plan": { "steps": [...] },
"pr": { "number": 456, "url": "..." }
}Source Preferences: {state-dir}/sources/preference.json - Caches task source selection:
{
"source": "custom",
"type": "cli",
"tool": "tea",
"savedAt": "2025-01-19T08:00:00.000Z"
}Core Workflow (Opus - Complex Tasks):
| Agent | Purpose |
|---|---|
| exploration-agent | Deep codebase analysis |
| planning-agent | Design implementation plans |
| implementation-agent | Execute plans with quality code |
| review-orchestrator | Multi-agent code review with iteration |
Quality Gates (Sonnet - Side Reviewers):
| Agent | Purpose |
|---|---|
| deslop-work | Clean AI slop from new work (committed but unpushed) |
| test-coverage-checker | Validate new work has test coverage |
| delivery-validator | Autonomous delivery validation (not manual) |
| docs-updater | Update docs related to changes |
Operational (Sonnet - Infrastructure):
| Agent | Purpose |
|---|---|
| task-discoverer | Find and prioritize tasks (multi-source) |
| worktree-manager | Create isolated worktrees |
| ci-monitor | Monitor CI/PR status with sleep loops |
| ci-fixer | Fix CI failures and review comments |
| simple-fixer | Execute predefined code fixes |
Reality Check (Opus - Plan Drift Detection):
| Agent | Purpose |
|---|---|
| plan-synthesizer | Deep semantic analysis with full context (opus) |
Data collection handled by JavaScript collectors (lib/reality-check/collectors.js)
awesome-slash/
|-- .claude-plugin/
| |-- marketplace.json # Claude Code marketplace manifest
|-- plugins/
| |-- next-task/ # Master workflow orchestrator
| | |-- commands/ # next-task, update-docs-around, delivery-approval
| | |-- agents/ # Specialist agents
| | |-- hooks/ # SubagentStop hooks for workflow automation
| |-- ship/ # PR workflow
| |-- deslop-around/ # AI slop cleanup
| |-- project-review/ # Multi-agent review
| |-- reality-check/ # Plan drift detection
|-- lib/
| |-- config/ # Configuration management
| |-- state/ # Workflow state management
| |-- sources/ # Multi-source task discovery
| |-- platform/ # Auto-detection
| |-- patterns/ # Code analysis patterns
| |-- utils/ # Shell escaping and context optimization
|-- mcp-server/ # Cross-platform MCP server
|-- scripts/install/ # Platform installers
|-- docs/
Required:
- Git
- Node.js 18+
Required for GitHub-backed workflows:
- GitHub CLI (
gh) with authentication
For Claude Code:
- Claude Code CLI
For OpenCode:
- OpenCode CLI (
opencode)
For Codex CLI:
- Codex CLI (
codex)
Contributions welcome. See CONTRIBUTING.md.
MIT - Avi Fenesh
- Issues: https://github.com/avifenesh/awesome-slash/issues
- Discussions: https://github.com/avifenesh/awesome-slash/discussions
Made with care for the AI coding community