OpenSpec 1.13.1 Hardens Its Spec-First Workflow for AI Coding Agents

AI coding assistants can write code quickly, but they also make assumptions quickly, and the intent behind a change tends to get lost once the chat session ends. OpenSpec, an open-source spec-driven development framework from Fission-AI, tackles that by having the developer and the AI agent agree in writing on what to build before any code gets written. The plan lives in the repository instead of in chat history.

How it works

A lightweight spec layer in markdown files describes how the system currently behaves. Each proposed change gets its own folder holding a proposal, design notes, a task list and “delta” specs that mark requirements as added, modified or removed instead of rewriting whole documents. Once the work ships, the archive command merges those deltas back into the main specs. Commands such as /opsx:explore, /opsx:propose and /opsx:apply walk an agent through exploring an idea, proposing a change and implementing it.

OpenSpec generates skills and slash commands for more than 30 AI tools, heise online reported when version 1.7.0 shipped. That release also added a machine-wide default store and a skip_specs: true flag so pure refactorings and documentation changes can be validated and archived without an artificial spec change.

Version 1.13.1: safer in unreviewed repos

Version 1.13.1, released Sept. 17, is meant to make OpenSpec safer in repositories you haven’t reviewed yet. A config.yaml value can no longer inject directives into agent instructions, crafted files can no longer hang openspec update or openspec archive, and a repository’s .npmrc can no longer redirect the update check. openspec status now ends with a “Next:” line naming the exact command that moves a change forward, and workflows check that a project has run openspec init before writing anything.

Updated Oct. 25, 2026: OpenSpec has since reached version 1.15.0, which adds a standalone installer that doesn’t need Node.js and an openspec update --dry-run option. The GitHub repository has more than 72,000 stars.