CODING AGENT

KittyCode

Hackable terminal coding agent. Minimal core, extensible skills, your way. A compact agent loop with local tools, context compression, and multi-provider support.

Design Philosophy

Minimal by Default

Intentionally small codebase. The core runtime is straightforward — agent loop, tools, context compression, CLI. Nothing extra.

Hackable by Design

Extend with skills, swap models, customize tools. The architecture is built to be modified, not just used.

Your Agent, Your Way

Drop-in skill system. Write a SKILL.md, add files, and your agent learns new capabilities instantly.

Multi-Provider

Works with both OpenAI-compatible and Anthropic APIs. Switch providers without changing your workflow.

Core Features

⚙️

Agent Loop with Parallel Execution

Minimal agent loop that supports parallel execution for multiple tool calls, keeping responses fast.

🧩

Skill System

Scan ~/.kittycode/skills for skill folders. Each SKILL.md is automatically discovered and injected into the agent's context.

📦

Built-in Tool Suite

Shell execution, file operations, web search, web fetch/crawling, TODO tracking, and sub-agents — all included.

💾

Session Persistence

Save and resume sessions across terminal restarts. Context compression keeps long sessions manageable.

⏸️

Cooperative Interruption

Press Esc to interrupt at the next safe checkpoint between LLM streaming, tool dispatch, and loop rounds.

Quick Start

# Install KittyCode
$ pip install kittycode
 
# Configure your API keys
$ kittycode --config
 
# Start coding
$ kittycode
 
# Run a one-shot prompt
$ kittycode -p "Explain the project structure"

Interactive Commands

/skills

List all loaded skills discovered at startup.

/model <name>

Switch between configured model providers on the fly.

/compact

Manually trigger context compression for long sessions.

/save & /sessions

Save current state and browse previous sessions.

Tech Stack

Python 3.10+ OpenAI API Anthropic API Terminal REPL SQLite Sessions Skill Discovery