Open Source Free
$0
Fully free. Bring your own API keys. Desktop app, CLI, and API. 15+ provider support.
Goose
The open-source AI agent that goes beyond code suggestions.
Goose is an open-source, extensible AI agent from Block (formerly Square) that runs locally and autonomously executes tasks: installs dependencies, edits files, runs tests, calls APIs, and browses the web. MCP-native from day one, it works with any LLM (Claude, GPT, Gemini, local models) and supports custom extensions via a simple YAML manifest. Crossed 30K+ GitHub stars since its public launch in January 2025.
+ Pros
- Fully local and Apache 2.0 licensed — no data sent to third parties beyond your chosen LLM
- MCP-native from day one — connects to any MCP server without extra configuration
- Provider-agnostic: works with Claude, GPT, Gemini, and local models via Ollama
- Simple YAML extension manifests make it easy to add and share custom capabilities
- 30K+ GitHub stars and active Block-backed community
− Cons
- Requires LLM API key setup — no built-in free inference tier
- No cloud-hosted option — all setup and maintenance is the user's responsibility
- Younger project — fewer built-in integrations than commercial tools like Cursor or Devin
- Desktop GUI currently macOS-only; Linux/Windows users are CLI-only
Pricing
Goose is Block's open-source autonomous AI agent, introduced in January 2025 as 'codename goose' and quickly becoming one of the most-starred local agent projects on GitHub. Unlike IDE-bound copilots, Goose runs entirely on your machine — it can install packages, edit files, execute shell commands, run tests, call APIs, and browse the web to gather information, all driven by a natural language instruction. It is provider-agnostic and MCP-native from day one, meaning it integrates with any Model Context Protocol server out of the box.
Extensions in Goose are defined as simple YAML manifests (a name, description, and a list of tools), making it easy for developers to package new capabilities and share them with the community. The agent maintains a persistent session with memory of previous steps, can run parallel sub-tasks, and supports all major LLM providers — Anthropic Claude, OpenAI GPT, Google Gemini, and local models via Ollama or LM Studio. A lightweight Desktop GUI was added for macOS users who prefer a visual interface without losing the full power of the CLI.
Key Features
• Fully autonomous local agent: shell, file editing, test running, web browsing, API calls
• MCP-native: integrates with any MCP server out of the box
• Any LLM: Claude, GPT, Gemini, local models via Ollama/LM Studio
• Extensible via YAML extension manifests — easy to build and share toolkits
• Persistent session memory across steps
• Parallel sub-task execution
• Desktop GUI for macOS (in addition to CLI)
• Apache 2.0 licensed — free to use, modify, and deploy
• 30K+ GitHub stars (May 2026)
Version History
TUI command on goose-cli, TUI diff viewer, goose review local code review, /goal self-evaluation command, hooks system, slash commands in ACP, Nushell support.
Hooks system, goose review, diff viewer, /goal command, slash commands in ACP, unified thinking effort control, GPT-5.5 support.
Build non-vulkan Linux variants using Ubuntu 22.04 for improved compatibility.
Hooks support, Agents CRUD, Projects as backend sources, auto-updating plugins, Linux Vulkan local inference, ACP streamable HTTP.
Bug fix: removed Windows CUDA build variant from release pipeline.
Signature Snippet
Run `goose session` and type: 'Clone the repo at github.com/me/api, install dependencies, run the test suite, identify the 3 failing tests, fix them, and open a PR.' Goose shells out to git, npm, the test runner, reads the failures, edits the source files, re-runs tests to verify, and uses the GitHub CLI to open the PR — all without leaving the terminal.More in this Space
Vix
Vix is a Go-native, open-source (AGPL-3.0) AI coding agent that slashes token costs by 40-50% using a stem agent architecture and Tree-sitter virtual filesystem. It rethinks the plan/execute loop — keeping LLM cache warm across Explore, Plan, and Execute phases — while shipping Programmable Workflows, Whiteboard Mode with voice AI, MCP server support, and a self-evolving agent that writes its own scheduled jobs and watchers.
Late — High-Leverage AI Agent Orchestration
Orchestrate an entire AI dev team on 5GB VRAM using ephemeral subagents, exact-match diffs, and a zero-dependency Go binary. Works with any OpenAI-compatible model — local or cloud.
Paca
Paca is a free, open-source, self-hosted Scrum board where AI agents work as equal teammates — assigned to sprints, picking up tasks, and collaborating on BDD specs alongside humans. Built as an alternative to Jira and Linear, it treats AI agents as first-class Scrum members.