SkillFoxx

OpenClacky CLI

OpenClacky is an open-source AI agent in Ruby with 16 tools, any OpenAI-compatible key and extension through skills.

  • 1,207 ★ on the repository
  • Last commit
  • MIT
CLI

Broad access

Why this level

  • The agent runs commands and edits files in your project, and server mode with address 0.0.0.0 exposes the interface to the network.
All reasons and checks

clacky-ai/openclackyAuthor:clacky-ai

Install

Works in

Manual install

gem install openclacky

The command from the Installation README section. Ruby 3.1 or newer is required. Start with `openclacky`.

Access and checks

How we review

Broad accessThe tool can change data, run commands or handle secrets.

  • The agent runs commands and edits files in your project, and server mode with address 0.0.0.0 exposes the interface to the network.

This is third-party code. Review the repository files before installing.

What it does

OpenClacky installs as the `openclacky` gem and runs with `openclacky` for an interactive session in the current folder or `openclacky server` for a web interface on port 7070. A model connects with a BYOK key from any OpenAI-compatible service. The agent keeps only 16 core tools and moves the rest into skills through the `invoke_skill` meta-tool. This shrinks the tool schema in every request. The README claims a near-perfect cache hit rate, automatic compression of long context while idle and skill self-evolution. By the README, token spend is about 0.8 of Claude Code on comparable tasks, with measurements done inside the project. The agent connects to Feishu, WeCom, WeChat, Discord and Telegram. There are installers for macOS and Windows, a script for Linux and a Docker image.

Who it is for. Developers looking for an open agent for code and terminal tasks with their own model and skill extensions.

Good fit when

  • You want a terminal agent with an MIT license and your own key instead of a subscription.
  • Token spend grows and you want an agent with a small tool set and a stable cache.
  • The agent must be connected to a team messenger such as Telegram or Discord.
  • Your workflows are packaged as skills and loaded through `invoke_skill`.

Not a fit when

  • You need verified cost measurements: README figures come from internal tasks and full reports are not yet published.
  • You require Claude Code with a built-in subscription and official support.
  • Ruby is unavailable and network-fetched installer scripts are not allowed.

Example request

Run `openclacky` in the project folder, connect a model key and ask it to add a settings page with tests to the app.

Limitations

Cost comparisons in the README come from the project internal tasks and the authors promise full reports later. The agent uses your OpenAI-compatible API key, so costs and quality depend on the chosen model. Installing the gem needs Ruby 3.1 or newer. On Linux and Windows the main install path runs a script downloaded over the network. The `server` mode with address `0.0.0.0` exposes the web interface to the network, and access is closed with the `CLACKY_ACCESS_KEY` key. The last commit dates from November 2026.

How to disable. Remove the gem with `gem uninstall openclacky`. For Docker stop and remove the `openclacky` container, data lives in the `openclacky-data` volume and in `~/.clacky`.

FAQ

How do I install OpenClacky?

Run `gem install openclacky`. macOS and Windows have `.dmg` and `.exe` installers, and Linux and macOS have the `scripts/install.sh` script from the repository.

How do I start the web interface?

With `openclacky server`: it opens on `http://localhost:7070` by default. Change the port with `--port` and the address with `--host`.

Which models work?

Any with an OpenAI-compatible API: connection by BYOK key, with no lock-in to one provider.

How do I extend the agent?

With skills: the agent calls them through `invoke_skill` and they load when needed instead of sitting as a schema in every request.

Can I run it in Docker?

Yes: `docker run -d --name openclacky -p 7070:7070 -v openclacky-data:/root/.clacky` with the registry image. Data stays in the volume.

Discussion

No questions or reviews yet.

CLI tools

A CLI agent swarm: a leader spawns workers in separate git worktrees, and they share tasks and exchange messages via files.

CLIBroad access1.5K
Editors’ pick

Open-source personal AI assistant on your own machine: answers in Telegram, Slack, Discord and WhatsApp, extended with skills and plugins

CLIBroad access394.4K

AgentOS runs a team of OpenClaw agents: workspaces, scheduled tasks, activity reports and human approvals

CLIBroad access134
Needs a VPN

A compact OpenClaw alternative: a personal assistant on Claude Agent SDK with agents in isolated containers and messengers added via skills

CLIBroad access31.1K

The clawctl CLI manages a fleet of AI agents on local machines over SSH and Ansible: install, start, upgrade and token tracking.

CLIBroad access67

A terminal AI agent in Go: Hub profiles, MCP servers, skills and a loop that retries until your check passes

CLIExtended access153

README in short

A Ruby package and agent server with console and web modes, model setup inside `openclacky` and messenger integrations. Docs are at openclacky.com.

README badge

Report a problem