gitmyhub

claude-engineer

Python ★ 11k updated 1y ago

Claude Engineer is an interactive command-line interface (CLI) that leverages the power of Anthropic's Claude-3.5-Sonnet model to assist with software development tasks.This framework enables Claude to generate and manage its own tools, continuously expanding its capabilities through conversation. Available both as a CLI and a modern web interface

A Python coding assistant powered by Claude 3.5 Sonnet that reads and edits your files, runs code, and can write new tools for itself mid-conversation, available as a terminal CLI or local web app.

PythonAnthropic APIsetup: moderatecomplexity 3/5

Claude Engineer v3 is a Python application that gives you a coding assistant powered by Anthropic's Claude 3.5 Sonnet model. You can use it either as a command-line tool in your terminal or through a web interface that runs locally in your browser. The main idea is that you have a conversation with the AI and it can help you write, read, and edit files, run code, manage packages, and perform other development tasks.

The distinguishing feature of this version is self-improvement: during a conversation, Claude can identify that it needs a new capability, write the code for a new tool itself, load it into memory, and start using it immediately. The project calls this dynamic tool creation, and it means the assistant becomes more capable the more you use it. Tools it creates are stored in a tools directory and can be reused in future sessions.

The web interface runs on a local server and includes image upload, markdown rendering, and a visual display of how many tokens the conversation has used so far. The command-line interface provides the same capabilities in a terminal, with text-based formatting and progress indicators. Both interfaces share the same underlying assistant logic.

Setup requires Python and an Anthropic API key. The README recommends installing a package manager called uv, then cloning the repository, creating a virtual environment, and running either app.py for the web interface or ce3.py for the CLI. The project also supports running Python code in a sandboxed environment via a separate service called E2B, which requires its own API key.

The built-in tools cover common development tasks: creating folders and files, reading multiple files at once, editing files, running a Python linter, and managing packages. The tool creator tool is what enables new tools to be added at runtime based on a natural language description during a conversation.

Where it fits