One command. Every language. Zero configuration.
Plus: A complete AI development methodology.
You're drowning in tools. Python needs black, ruff, pytest, mypy. Rust
wants cargo fmt, clippy, cargo test. Node.js demands prettier, eslint,
jest. Your .gitignore is longer than your actual code.
But there's a bigger problem: Your AI assistant is brilliant but chaotic. One moment it's over-engineering, the next it's missing critical security checks. No consistency. No methodology. No quality guarantees.
Khive solves both problems:
- 🔧 Unified Tools: One command for all languages, all workflows
- 🧠 AI Framework: Transform any AI into a structured development team
# The tools you know and love
khive init # Sets up any project in seconds
khive fmt # Formats everything perfectly
khive ci # Tests across all languages
# The AI revolution (via Roo/Cline integration)
# Your AI now has 6 specialized roles following proven workflowsKhive provides carefully crafted prompts that transform your AI assistant into six specialized experts:
| Role | Specialist | Expertise | Key Responsibilities |
|---|---|---|---|
| 🎼 | Orchestrator | Project Management | Coordinates workflow, enforces quality gates, manages GitHub issues/PRs |
| 🔭 | Researcher | Technical Analysis | Investigates solutions, compares approaches, provides evidence-based recommendations |
| 📐 | Architect | System Design | Creates technical specifications, designs scalable architectures |
| 🛠️ | Implementer | Development | Writes production code following TDD, achieves >80% test coverage |
| 🩻 | Reviewer | Quality Assurance | Ensures spec compliance, security, performance, code quality |
| 📚 | Documenter | Knowledge Management | Creates user guides, API docs, maintains documentation |
# 1. Install Khive to get the prompt system
pip install khive[all]
khive init
# 2. Your AI assistant (via Roo) can now access specialized roles
# In Roo: "As @khive-orchestrator, plan the OAuth implementation"
# In Roo: "As @khive-researcher, compare JWT vs session auth"
# In Roo: "As @khive-implementer, build the auth service"
# 3. Each role follows the Golden Path workflow
Research → Design → Implement → Review → Document → MergeEvery Khive AI role follows structured workflows with quality gates:
📋 Templates for Every Stage
├── RR (Research Reports) - Evidence-based analysis
├── TDS (Technical Design Specs) - Complete blueprints
├── IP (Implementation Plans) - TDD-first development
├── TI (Test Implementation) - Comprehensive testing
├── CRR (Code Review Reports) - Quality verification
└── Documentation - User-facing guides
Quality Gates:
- ✅ Research must cite sources (search IDs)
- ✅ Design must reference research
- ✅ Code must have >80% test coverage
- ✅ Reviews verify spec compliance
- ✅ Docs required before merge
# Install Khive and set up your project
pip install khive[all]
cd your-project
khive init
# Now you have:
# 1. ✅ All your tools configured and working
# 2. ✅ .khive/prompts/ with AI team roles
# 3. ✅ Templates for structured development
# 4. ✅ Integration with Roo for AI assistanceBefore we dive into AI, let's not forget Khive's powerful unified tooling:
# One command for everything
khive fmt # Format Python, Rust, TypeScript, Markdown
khive ci # Run all tests in parallel
khive commit # Smart commits with conventional format
khive pr # Create PRs from terminal
khive clean # Manage branches intelligently
# Extensible with your workflows
echo '#!/bin/bash
# Your custom logic
' > .khive/scripts/khive_deploy.sh
khive deploy # Now everyone has your deployment flowInstead of one general AI, you get six experts. The Architect thinks differently than the Implementer. The Researcher provides evidence, not opinions.
The Golden Path isn't arbitrary - it's based on successful enterprise development practices, encoded into prompts.
Every stage has checks. No more "LGTM" reviews. No more missing tests. No more outdated docs.
# AI roles use Khive tools naturally
@khive-researcher: "I'll search for solutions" → khive info search
@khive-implementer: "Running tests" → khive ci
@khive-reviewer: "Checking formatting" → khive fmt --check| Metric | Without Khive | With Khive | Impact |
|---|---|---|---|
| Project Setup | 2-4 hours | 2 minutes | 99% faster |
| AI Development Quality | Inconsistent | Structured & Verified | Predictable |
| Test Coverage | "Sometimes" | Always >80% | Reliable |
| Documentation | "TODO" | Always Current | Professional |
| Code Reviews | Superficial | Comprehensive | Secure |
- Tools should unify, not multiply - One interface for everything
- AI needs structure - Roles, workflows, and quality gates
- Humans lead, AI executes - You set direction, AI handles details
- Evidence over opinion - Every decision traced to research
- Quality is non-negotiable - Tests, reviews, and docs required
- Unified CLI for all tools
- AI team prompts and roles
- Golden Path methodology
- Roo/Cline integration
- Custom script extensions
- MCP server support
- Native orchestration CLI (
khive orchestrate) - VS Code extension
- Cloud-based AI team coordination
- Template marketplace
- Autonomous PR workflows
# Get started in 60 seconds
pip install khive[all]
khive init
# Explore the AI prompts
ls .khive/prompts/roo_rules/
# Start using structured AI development
# (In your AI assistant via Roo)"Khive turned my ChatGPT from a code monkey into a senior engineering team." - Staff Engineer, FAANG
"The combination of unified tools and AI methodology is genius. Ship faster with higher quality." - CTO, YC Startup
"Finally, AI that follows our standards instead of making them up." - Engineering Manager, Fortune 500
- Golden Path - The methodology
- Discord - Join the community
Apache 2.0 - Because great tools should be free.
🐝 Khive: Where tools become unified. Where AI becomes structured.
⭐ Star us on GitHub • 📦 Install from PyPI
Built by developers who believe AI should amplify expertise, not replace it.