OpenSpec 🤖 Agent Free

What it does

OpenSpec adds a lightweight specification layer between you and your coding agent. Before any code is written, each change is written up as a small set of documents that you and the agent review and agree on; once the work is done, the change is archived and its requirements are merged into the project's living specs. It is iterative rather than phase-gated and designed for existing codebases as much as new ones.

What is inside / how it works

Running openspec init creates an openspec/ folder with specs/ for the current requirements and changes/ for work in progress. The default workflow uses a few agent commands: /opsx:explore to think through an idea against your code, /opsx:propose to create a change folder with a proposal, requirement specs with concrete scenarios, a design note and a task checklist, /opsx:apply to implement the tasks and /opsx:archive to fold the result back into the specs. An expanded profile adds steps such as verify, fast-forward and bulk archive. Specs are plain Markdown, and any artifact can be edited at any time. A beta feature called Stores keeps planning in a separate repository so several code repos and teams can share one set of specs. The CLI collects anonymous command-usage telemetry, which can be switched off.

Works with

The supported-tools page lists more than 30 integrations, including Claude Code, Cursor, GitHub Copilot, Codex, Gemini CLI, OpenCode, Cline, Kilo Code, Kiro, Continue, Amazon Q Developer, Qwen Code, Antigravity and Devin Desktop (formerly Windsurf).

How to install or start

OpenSpec needs Node.js 20.19.0 or later. Install the CLI globally, then initialise your project and pick your tools:

npm install -g @fission-ai/openspec@latest
cd your-project
openspec init

Run openspec update after upgrading to refresh the agent instructions.

Pricing

Free and open source under the MIT licence.

Who should use it

Developers who want the agent to agree on a plan before it edits code, without adopting a heavy process. It suits ongoing work on existing projects well. It works best with strong reasoning models and a clean context window, and teams need to keep reviewing the generated specs for them to stay useful.