Skip to main content

Boring Vibecoder Assistant: An AI-powered development support tool handling project planning, logic verification, and knowledge management. Works alongside Cursor, Copilot, and others to ensure rigorous, efficient workflows.

Project description

PyPI version License Vibe Coder

Boring for Gemini

Autonomous AI Agent Loop with VibeCoder Experience

English | 繁體中文


✨ The Vibe Coder Experience (V10.24)

No Code Needed. Just describe the vibe.

Boring-Gemini now features a Universal Natural Language Router. You don't need to remember 98+ complex tools. Just say what you want in English or Chinese:

"Search for authentication logic"

"Review my code for security issues"

"幫我寫測試" (Help me write tests)

"我想做登入功能" (I want to build login feature)

Try it in your terminal:

boring-route "review my code"
# 🎯 Routes to boring_code_review (100%)

👉 Learn more about Vibe Coder Experience


⚡ boring 的厲害 (Why Boring?)

Boring isn't just an MCP server; it's a Intelligence Maximization System:

  1. 🧠 Autonomous Loop: Not just a chatbot. Boring runs in a loop (boring start), thinking, coding, testing, and fixing until the job is done.
  2. 🕵️ Hybrid RAG: Advanced code search combining keywords, vectors, and dependency graphs (HyDE + Cross-Encoder). It finds code you didn't even know existed.
  3. 🛡️ Security Shadow Mode: Safe execution sandbox. It catches dangerous operations before they happen.
  4. ⚡ 30% Faster: Smart caching and optimized router reduce context usage by 80% (from 98 tools to 19).
  5. 🧩 Vibe Coder: The most human-friendly AI coding interface. Zero friction between thought and code.

🚀 Quick Start

Downloads smithery badge

Boring: Your Autonomous Coding Partner

Enterprise-grade Autonomous AI Development Agent
A full-language automated coding and verification engine built for Cursor / Claude Desktop / VS Code / Gemini CLI.

中文版 README | Full Documentation


🚀 Core Advantages

Feature Description
🌐 Polyglot & CLI Native Seamless switching between Gemini CLI and Claude Code CLI, zero API key required
🛡️ Parallel Verification Multi-threaded parallel verification, 3-5x performance boost
🧠 RAG Memory Hybrid Search (Vector + Keyword) + dependency graph for real-time retrieval
🛡️ Shadow Mode High-risk operations require human approval, with persistent config
📐 Spec-Driven 100% specification consistency from PRD to Code
🔒 Quality Gates CI/CD multi-tier gates + multi-language linting + 20+ file type security scanning
Vibe Coder Pro Doc Gen

[!IMPORTANT] Boring is now primarily an MCP tool (used via Cursor / Claude Desktop / IDE)

  • Not recommended to run boring start directly in CMD/terminal: Gemini CLI no longer supports free authorization (unless using API, which is not well-tested)
  • Recommended usage: Use Boring tools through Smithery or MCP config in your IDE or Client
  • Monitoring tools still work: boring-monitor, boring-dashboard can be run locally

Most features are optimized for the MCP environment. CLI mode is no longer the primary supported method.


📦 Quick Installation

Option 1: Smithery (✅ Recommended)

npx -y @smithery/cli@latest install boring/boring --client gemini-cli

⚠️ Gemini Client Users: If you encounter issues installing via Smithery, please use Option 2 (Local pip). Direct Smithery integration on Gemini Client can be intermittent.

Option 2: Local pip Installation

# Install with all features (Recommended for Vibe Coder)
pip install "boring-aicoding[all]"

# Or minimal install
pip install boring-aicoding

🤔 Which one should I choose?

Feature [all] (Full / Local) Lite (Basic / Smithery Default)
RAG Memory ✅ Vector + Semantic ⚠️ Keyword only (No vector DB)
Self-Verify ✅ Can run tests (boring verify) ❌ Cannot verify (Missing pytest)
Dashboard ✅ GUI Available ❌ None
Vibe Coding Full Experience (Think + Fix) ⚠️ Lite (Write code only)

Option 3: Clone from GitHub (Fallback)

Best for: Developers or if pip install fails

git clone https://github.com/Boring206/boring-gemini.git
cd boring-gemini
pip install -e .

⚙️ MCP Configuration

For Smithery

{
  "mcpServers": {
    "boring": {
      "command": "npx",
      "args": ["-y", "@smithery/cli", "run", "@boring/boring", "--config", "{}"]
    }
  }
}

For Local pip

{
  "mcpServers": {
    "boring": {
      "command": "python",
      "args": ["-m", "boring.mcp.server"],
      "env": {
        "BORING_MCP_MODE": "1",
        "PROJECT_ROOT_DEFAULT": ".",
        "BORING_MCP_PROFILE": "lite"
      }
    }
  }
}

🎛️ Tool Profiles (V10.24)

Problem: 98 tools can overwhelm LLM context windows. Solution: Choose a profile to expose only the tools you need.

Profile Tools Best For
minimal 8 Simple workflows, minimal context
lite 20 Recommended - everyday development
standard 50 Full-featured projects
full 98+ Power users who need everything

Configure in .boring.toml:

[boring.mcp]
profile = "lite"  # Options: minimal, lite, standard, full

Or via environment variable:

export BORING_MCP_PROFILE=lite

🎯 Universal Router: With lite profile, use boring("search for auth code") - the router automatically directs to the right tool!


🎯 Quick Start Prompts

Prompt Usage
/vibe_start Start a new project with AI guidance
/full_stack_dev Build a complete full-stack application
/release-prep Turbo: Auto-bump version & git tag
/quick_fix Auto-fix all linting and formatting errors
/smart_commit Generate semantic commit messages

📚 Documentation

Category Links
Getting Started Vibe Coder Guide · 🗣️ Natural Language Prompts · Quick Tutorials
Features MCP Tools (55+) · Shadow Mode · Quality Gates · Monitoring
Guides Cookbook · Pro Tips · Git Hooks · Workflows
Learning Tutorials · Skills Guide · Knowledge Mgmt
Advanced Plugins · API Integration · Human Alignment
Reference Architecture · Security & Privacy · Agent Comparison · V10 Changelog

✨ Vibe Coder Pro Toolset

🎉 Interactive Tutorial on Startup! First-time MCP connection triggers mcp_intro to guide new users.

🗣️ Natural Language Triggers (No Code Required!)

Just say these to Boring:

Goal Just Say
Write Tests Help me write tests for auth.py
Review Code Review my code
Health Check Vibe Check my project
Check Impact Check impact of utils.py
Plan Feature I want to add login, Plan this feature

👉 Full Prompt Guide

🧰 Core Tools

Tool Description Example
🧪 boring_vibe_check Project Health Scan - Coverage, Docs, Security (S-F Grade) boring_vibe_check(project_path=".")
📊 boring_impact_check Impact Analysis - Trace reverse dependencies (L1/L2/L3) boring_impact_check(file_path="core.py")
🧪 boring_test_gen Smart Test Gen - Auto-generate simple tests (Py/JS/TS) boring_test_gen(file_path="utils.py")
📝 boring_review Friendly Review - Constructive feedback & suggestions boring_review(file_path="app.py")
🚀 boring_perf Perf Tips - Identify bottlenecks boring_perf(file_path="main.py")
📐 boring_arch Arch Check - Visualize module structure boring_arch(project_path=".")
📄 boring_doc_gen Doc Gen - Auto-generate docstrings boring_doc_gen(file_path="api.py")

🔐 Security Scan Integration

Vibe Coder Pro includes built-in Token detection for:

  • Python: PyPI, AWS, GCP, Azure, etc.
  • JS/Node.js: NPM, Vercel, Supabase, Firebase, Stripe.
  • General: GitHub, GitLab, Slack, SendGrid (20+ types).
# Health check triggers security scan automatically
result = boring_vibe_check(project_path=".", max_files=100)
print(result["security_issues"])

🧠 External Intelligence

Boring integrates the most powerful external MCP tools by default, turning the Agent into a super engineer.

Tool Function How to Use
Context7 📚 Real-time Docs
Query latest library usage, solving stale training data issues.
context7_query_docs
Sequential Thinking 🤔 Deep Thinking
Force Agent into a full chain of thought from analysis to verification before coding.
sequentialthinking
Critical Thinking 🧐 Critical Thinking
Self-reflection and blind spot detection for high-quality Code Review.
boring-route "think deeper"
Boring Monitor 🖥️ TUI Dashboard
Real-time terminal view of status, logs (v10.23+).
boring-monitor / python -m boring.monitor
Boring Dashboard 🎨 GUI Dashboard
Comprehensive web view with Brain Explorer.
boring-dashboard / python -m boring dashboard / python -m boring.monitor --web

🚀 Performance (v10.21.0)

  • Thread-local SQLite: Zero-overhead DB connections.
  • WAL Mode: 50% faster concurrent reads.
  • Smart Caching: 30s query cache & Pattern cache for instant RAG.

🛡️ Shadow Mode

Shadow mode protects you from destructive AI operations:

DISABLED  ⚠️  No protection (Container only)
ENABLED   🛡️  Auto-approve safe, Block risky (Default)
STRICT    🔒  Approve all writes (Production)
boring_shadow_mode(action="set_level", level="STRICT")

🎯 Future Vision

Note: The following features require server support (not yet implemented)

  • 🌐 Boring Cloud: Cloud collaboration and team sharing
  • 🤝 Team Workflows: Multi-person workflow synchronization
  • 🔐 Enterprise SSO: Enterprise-grade identity authentication

🙏 Acknowledgments


📄 License

MIT License - Open source and free to use


🔗 Links

GitHub PyPI Smithery

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

boring_aicoding-10.24.4.tar.gz (399.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

boring_aicoding-10.24.4-py3-none-any.whl (467.1 kB view details)

Uploaded Python 3

File details

Details for the file boring_aicoding-10.24.4.tar.gz.

File metadata

  • Download URL: boring_aicoding-10.24.4.tar.gz
  • Upload date:
  • Size: 399.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.11.14

File hashes

Hashes for boring_aicoding-10.24.4.tar.gz
Algorithm Hash digest
SHA256 d36d301dfc49a4ba40a815d0c2159c5b74cfc09fd54a2a368b22fdf287b05295
MD5 f62c8eff4f1ce047598692d6228c1320
BLAKE2b-256 f3b24a6932ccef27a70a21d18374e27daf8e53ad2813b833fbe79d3d48c6ce5e

See more details on using hashes here.

File details

Details for the file boring_aicoding-10.24.4-py3-none-any.whl.

File metadata

File hashes

Hashes for boring_aicoding-10.24.4-py3-none-any.whl
Algorithm Hash digest
SHA256 effbadcd793cb2e86cbc6576335cd0e5473bfeb6ed6bbbca2bd73e84ce9bb542
MD5 e35129c7adfca3980cb6a5373d7c7c19
BLAKE2b-256 6191d3b5d23ad17896e976d3a41d65d66253b76060c5b78b9d398f801d03d7fd

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page