Spec Kitty CLI Enhances AI Code Development Workflow

What changed

Spec Kitty, an open-source command-line interface (CLI) tool, has released a prerelease version, v3.2.0rc37, for its Team Kitty 4.x line, with a specific release candidate, 4.0.0rc2, available for testing. This tool is designed to facilitate spec-driven development for AI coding agents and multi-agent workflows, aiming to create governed software factories. The core functionality revolves around transforming product intent into a repository-native AI coding workflow, following a sequence of spec -> plan -> tasks -> next -> review -> accept -> merge. Spec Kitty integrates with various AI coding agents, including Claude Code, Codex, Cursor, Gemini, GitHub Copilot, and Windsurf, by providing isolated Git worktrees for parallel implementation without causing branch chaos. It ensures that specifications, plans, work packages, acceptance criteria, review states, and merge decisions are stored directly within the repository, serving as the source of truth. The tool emphasizes a transparent development process, where humans define intent and acceptance criteria, agents implement within traceable worktrees, and reviewers manage the acceptance and merge process with an audit trail. The goal is to establish a durable operating system for agentic coding rather than simply generating more prompt text.

Why it matters for builders

For developers working with AI coding assistants, Spec Kitty offers a structured approach to manage complex projects. It helps prevent the loss of requirements, decisions, and acceptance criteria that can occur in ad hoc AI coding sessions. By storing all development artifacts in Git, it provides a clear audit trail and a more organized workflow, especially when multiple agents or developers are involved. The tool's ability to isolate agent work in separate Git worktrees is crucial for managing parallel development efforts without conflicts.

Practical impact

Builders can adopt Spec Kitty to transition from less structured "vibe coding" to a more repeatable, spec-driven development process. The CLI offers commands for initializing projects (spec-kitty init), verifying setup (spec-kitty verify-setup), opening a local Kanban dashboard (spec-kitty dashboard), and managing the agent's next actions (spec-kitty next). It supports integrating AI coding agents through slash commands or skills. Common use cases include turning GitHub issues or bug reports into executable work packages, coordinating multiple AI coding agents, and ensuring architectural decisions remain close to the code. The tool also provides governance layers, with commands like spec-kitty dispatch to load governance context before an agent performs work. Installation is recommended via pipx install spec-kitty-cli for environment isolation.

Caveats and source limits

The current release mentioned is a release candidate (v3.2.0rc37 and 4.0.0rc2), indicating that the stable launch acceptance is pending. While the tool supports a wide range of AI coding agents, the specific performance or integration nuances with each agent are not detailed in the provided excerpts. Pricing information for any potential hosted services or advanced features is not available, as the tool is described as local-first with optional hosted integrations. The excerpt focuses on the CLI's functionality and philosophy, with deeper dives into specific technical implementations like the trail model or host-surface parity requiring consultation of linked documentation.