k4000

Crates.iok4000
lib.rsk4000
version0.3.12
created_at2025-11-28 11:28:26.079948+00
updated_at2026-01-23 10:55:48.553434+00
descriptionAI-assisted feature development and code review MCP servers
homepage
repositoryhttps://github.com/yourusername/k4000
max_upload_size
id1955173
size621,992
Camilo Tapia (camme)

documentation

README

Koder 4000 (K4000)

AI-assisted feature development and code review system using MCP (Model Context Protocol).

What is K4000?

K4000 helps LLMs create well-defined software features through a structured workflow:

  1. Feature Analysis - Takes a feature description, analyzes it for gaps, and generates clarifying questions
  2. Q&A Session - Humans answer questions to refine the feature specification
  3. Implementation - Following TDD principles, tests and code are written
  4. Code Review - Automated review from 5 perspectives with iterative feedback

How to install and use

cargo install k4000
cd [your project]
claude mcp add --transport stdio k4000-feature -- ~/.cargo/bin/k4000-feature-mcp
claude mcp add --transport stdio k4000-review  -- ~/.cargo/bin/k4000-review-mcp

Architecture

K4000 consists of two MCP servers:

  • Feature Analyzer MCP - Manages feature documents, questions, and status tracking
  • Code Review MCP - Orchestrates code review with specialized perspectives

Both servers communicate via the Model Context Protocol, allowing any MCP-compatible client (like Claude Code) to use them.

Quick Start

Build

cargo build --release

Configure MCP Client

Add to your MCP configuration:

{
  "k4000-feature": {
    "command": "./target/release/k4000-feature-mcp",
    "env": { "K4000_DATA_DIR": "./data" }
  },
  "k4000-review": {
    "command": "./target/release/k4000-review-mcp",
    "env": { "K4000_DATA_DIR": "./data" }
  }
}

Usage

See CLAUDE.md for detailed workflow instructions and tool documentation.

Project Structure

k4000/
├── Cargo.toml                 # Workspace configuration
├── CLAUDE.md                  # Detailed usage documentation
├── crates/
│   ├── k4000-core/           # Shared types and storage
│   │   └── src/
│   │       ├── lib.rs
│   │       ├── error.rs      # Error types
│   │       ├── storage.rs    # File-based state management
│   │       └── types/
│   │           ├── feature.rs # Feature types
│   │           └── review.rs  # Review types
│   ├── k4000-feature-mcp/    # Feature Analyzer MCP server
│   │   └── src/
│   │       ├── main.rs
│   │       └── server.rs
│   └── k4000-review-mcp/     # Code Review MCP server
│       └── src/
│           ├── main.rs
│           └── server.rs
└── data/                      # Runtime data (gitignored)

Review Perspectives

The Code Review MCP analyzes code from 5 perspectives:

Perspective Focus
Bugs Logic errors, edge cases, null handling
Best Practices Coding standards, patterns, maintainability
Security Vulnerabilities, injection, authentication
Performance Efficiency, memory, algorithmic complexity
Simplification Code that could be cleaner or simpler

Feature Status Workflow

draft → analyzing → awaiting_answers → ready → implementing → reviewing → done

License

MIT

Commit count: 0

cargo fmt