Skip to content
ReviewRamble

The weekly brief

AI, reviewed, in your inbox.

  • New reviews of tools, models, MCP servers and repos, with the verdict up front
  • Re-checked prices and ratings when something changes
  • New analysis pieces as they are published

Models, agents, MCP servers and the tools actually worth your time.

One email a week. Unsubscribe with the link in any brief. How we use your email

Spec-driven development

Spec-driven development is a way of working with coding agents in which you first write a specification of what to build and why, then have the agent turn it into a plan, a task list and finally code.

Also called: spec-first

By Divya Sharma · Checked Nov 15, 2026 · How we review

Asking an agent to "build a login page" leaves it to guess the details. Spec-driven development moves the decisions up front. The spec, written and reviewed by a person, becomes the source the agent works from.

How it works

A typical flow has a few stages, each reviewed before the next:

1. Specify: describe the feature, its users and the requirements, without choosing the technology. 2. Plan: the agent proposes a technical design that fits the codebase. 3. Tasks: the plan is split into small, checkable steps. 4. Implement: the coding agent works through the tasks.

GitHub's open-source Spec Kit packages this as a set of commands run in the agent's chat, with a project constitution that sets rules for every feature. Kiro builds a similar flow into its editor as specs made of requirements, design and tasks.

How it differs from vibe coding

Vibe coding skips review and lets the code speak for itself. Spec-driven development puts review at each stage, so mistakes are caught in the plan rather than in a large diff.

Limits

Writing a good spec takes time, and for small changes the overhead is not worth it. A spec can also drift from the code if it is not updated. Project rule files, covered under AI coding rules file, hold the standing guidance that applies to every spec.

Related terms

Coding agent
A coding agent is an AI agent that works on software directly: it reads a codebase, edits files, runs commands and tests, and repeats until a change works, rather than only suggesting code for a person to paste.
AI code editor
An AI code editor is a code editor or IDE with language models built in, offering completions, chat about the codebase and an agent mode that can plan and make changes across many files.
Vibe coding
Vibe coding is building software by describing what you want to an AI model and accepting the code it writes without reading or reviewing it, judging the result only by whether the program seems to work.
Agent instructions file (AGENTS.md)
An agent instructions file is a Markdown file in a code repository, such as AGENTS.md or CLAUDE.md, that a coding agent reads at the start of a session to learn the project's commands, conventions and rules.

Sources