CheetahClaws CLI
A CLI agent that works with project files and shell commands, lets you choose a model provider and connects external MCP tools.
- 796 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- The agent executes commands with user permissions, so a mistaken instruction can overwrite or delete project files.
- A cloud provider receives supplied project context and consumes account funds.
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/cheetahclawsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent executes commands with user permissions, so a mistaken instruction can overwrite or delete project files.
- A cloud provider receives supplied project context and consumes account funds.
This is third-party code. Review the repository files before installing.
What it does
CheetahClaws runs a local coding agent with a model chosen through an API or local server. It reads and edits files, executes commands and saves conversations for resumption. MCP servers and plugins add tools. You can change models and work modes in the terminal without rebuilding the application. It supports planning, task dependencies and Markdown skills. A web interface is also available, but the Python-package CLI is sufficient to begin.
Who it is for. For developers who want a configurable terminal agent using an external or local model.
Good fit when
- Try the same task with different available models.
- Use a local provider with a compatible API.
- Resume a saved conversation after the process stops.
- Give the agent tools from an existing MCP server.
Not a fit when
- A program must not execute commands with the user’s permissions.
- You need every action to prompt for approval by default.
- Source code must not reach the selected cloud provider.
Example request
Start CheetahClaws with approval prompts and have it inspect the test project structure before making changes.Limitations
The CLI needs Python 3.10 or later; check the interpreter version before installing the package. A cloud provider needs a key and reachable API; local use needs a model server already running. Automatic approval is enabled by default, so use `--ask` for the first run or save approval prompts with `/auto off`. Image and speech features require optional dependencies; install those separately when the chosen task needs them.
How to disable. Exit the CLI and stop any separately running web processes. Inspect saved conversations and provider configuration before deleting local state.
FAQ
How do I resume the last conversation?
Use `/resume last`. `/resume` without an argument lists saved sessions by name.
Which MCP transports are supported?
stdio, SSE and HTTP. Connected tools join the agent’s tool set.
Do I need to build a Node.js client?
No. The CLI installs as a Python package without a separate Node.js client build.
How do I connect a local Git MCP server?
Run `/mcp add git uvx mcp-server-git` inside the agent. `/mcp reload` reloads the configuration.
Can I avoid a cloud model?
Yes. Ollama, LM Studio and compatible local endpoints are supported. Prepare the server and an appropriate model separately.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI connects installed coding agents into a team, dispatches work between sessions and collects results across machines.
Open-source coding agent for VS Code, JetBrains, the terminal and desktop, with any model and MCP support
A tmux-based TUI that runs Claude Code, Codex and other agents in one list: status, quick prompts, worktrees and diff review
A terminal agent explores projects, prepares plans and makes edits or runs commands under a chosen approval policy.
A desktop app with a local MCP server that lets ChatGPT work with files, commands, skills and other MCP servers
One runtime for Claude Code, Codex, OpenCode and more: a run counts as done only when checks the agent did not write pass
README in short
The project saves conversations after every turn using atomic writes. Plugin authoring guides cover additional tools, commands and skills.