Spec Kitty Launches to Bring Structure, Memory, and Governance to AI Development Teams

Spec Kitty is an open-source command-line tool (MIT) that works inside the AI coding agents your team already uses, such as Claude Code, Codex, Cursor and Copilot. Before any code is written it asks what to build and how, writes a spec and plan you review, runs agents on the work in parallel git worktrees, and has a separate reviewer check each piece against the spec. Every question, answer and review is saved as files in your repo. Install it with pipx install spec-kitty-cli.

Instead of a chat prompt that disappears when the session closes, each piece of work starts as a written spec and plan your team can read and change, and ends with a review against that spec. The decisions made along the way stay in the repository next to the code.

Why Spec Kitty?

A coding agent handed a rough request fills the gaps itself, and you find its guesses in review. Spec Kitty does not replace the agent. It adds commands to it that ask the open questions first, write the answers down, and give the agent a reviewed spec to build from.

Step 1: Specification Planning

Begin by defining the intended goals, context, expected behaviors, risks, and acceptance criteria. Creating a detailed spec statement enhances the alignment within the team and offers a clear directive for both human and AI contributors.

Step 2: Work Package Breakdown

Transform ideas and project intents into manageable work packages. This systematic division allows for comprehensive understanding and distribution of tasks, ensuring each unit aligns with the project's primary objectives.

Step 3: Implementation Guidance

With Spec Kitty, AI agents receive precise instructions derived from well-constructed specs. This enhances the code-generation process, promoting efficiency and accuracy in execution.

Step 4: Review Trails

Establish a review trail that makes checking compliance against the spec effortless. This step ensures that any output is not only consistent with initial specifications but also transparent for review by stakeholders.

Step 5: Decision History Recording

Keep a record of all decision-making processes involved in development. This historical log aids in understanding the *why* behind actions, which is essential for future reference and governance.

Step 6: Project Memory Conservation

Preserve project memory to enhance continuity and efficiency in future endeavors. Spec Kitty helps in archiving intents and decisions, making this context readily available for both current and new team members.

Frequently Asked Questions

What is Spec Kitty?

Spec Kitty is an open-source command-line tool that runs your AI coding agents through spec, plan, build and review, and saves every decision as files in your repo. Install with pipx install spec-kitty-cli.

How does Spec Kitty solve workflow issues?

It asks what to build before any code, writes the answers into a spec and plan your team reviews, and keeps that record in the repo so the next person (or agent) starts with the context.

Can Spec Kitty replace AI coding tools?

No. It runs inside the coding agents you already use, on your own model keys.

How does Spec Kitty keep agent work on track?

A reviewer that is not the agent that wrote the code checks each piece of work against the spec's acceptance criteria before it merges.

Who benefits from using Spec Kitty?

Engineering teams that already build with AI coding agents and want one way of working with them.