Atelier plugin
Skills for spec-driven development: brainstorming, planning, debate-style review and idea grilling before the first commit.
- 51 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Subagents launch automatically, read code and edit project files during implementation.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/atelierDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Subagents launch automatically, read code and edit project files during implementation.
This is third-party code. Review the repository files before installing.
What it does
Atelier guides an agent from an idea to verified code and keeps decisions with the developer. The `atelier-orchestrator` skill picks the process size. A small task gets a short plan in the conversation. A large one goes through `spec-brainstorm` with a `design.md`, `spec-plan` with an approved `plan.json`, and `spec-implement`, where tasks with a settled design go to subagents. `oracle-grill-me` asks questions one at a time until key decisions are explicit. `code-review` runs review as a debate: a scout triages the diff, critics raise findings, an author defends the change and a judge rules only on what stays contested. There are skills for commits, pull requests, handoff and debugging. Every skill stops after its step and you start the next one with a separate request.
Who it is for. Developers who run large changes in Claude Code, Codex or Cursor and want design and plan on record before code.
Good fit when
- A migration or large refactor needs a `design.md` and an approved plan before implementation.
- A proposal feels settled too quickly and needs a grilling.
- A branch diff should be reviewed by several critics with an author defense.
- An unattended run of a task plan should go in batches with a review after each.
Not a fit when
- A one-line fix where the process is excessive.
- You want a fully autonomous agent with no approval stops.
- The environment lacks subagents while review relies on parallel critics.
Example request
Grill me with oracle-grill-me on this database migration plan, then write a spec-plan.Limitations
The project calls itself a personal developer toolkit, so its decisions reflect the author taste. Skills split work into separate requests and each one stops after a step, which slows small tasks. Review and planning launch parallel subagents, which raises token use. For OpenCode the skill commands are created at install, so the order of initialization and skill installation matters.
How to disable. Remove the skills through `npx skills` or by deleting folders from the agent skills directory, and undo the settings that `init` created for your harness. `design.md` and `plan.json` files in projects are deleted manually.
FAQ
How do I install the Atelier skills?
Run `npx skills add martinffx/atelier`. Update with `npx skills update martinffx/atelier`.
What is the init command for?
`npx @martinffx/atelier@latest init --harness` configures the chosen harness, such as Claude Code, Codex, Cursor or OpenCode.
Why do skills stop after each step?
This records decisions in `design.md` and `plan.json` and ties implementation to an approved plan.
How does review work?
`code-review` runs critics in parallel, the author answers each finding and a judge rules only on what stays contested.
Is there a shortcut for small tasks?
Yes. For bounded work the orchestrator gives a short plan in the conversation and implements it directly after approval.
Discussion
No questions or reviews yet.
Related
PluginsA plugin with three skills for large development work: project analysis, phased tasks tracked as GitHub Issues, parallel agents and code review
A skill that interviews you and writes requirements.md, design.md and tasks.md so Claude Code and Cursor work from one plan
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Orchestrator skill: from an idea the agent writes requirements and a spec, splits work into tasks, builds the project, checks it
A CLI that guides AI coding from idea to review through requirements, plan, code and checks and keeps traces in the repo.
Nine JS Mastery skills take a change through phases: from idea and spec to verification, tests, release texts and a docs sync
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Atelier (this card) | no | 51 | |
| OpenSpec | no | 74,784 | keeps change specs inside the project |
| Superpowers | no | 316,679 | a general development process skill set |
README in short
The project describes a flow from request through design and plan to implementation, review and pull request. Decisions live in files and each step needs a separate launch.
README badge
Original SKILL.md
--- name: atelier-orchestrator description: > Skill routing and workflow orchestration. Selects Inline Plan or Spec-backed Plan, routes to the correct workflow skill, and manages transitions between phases. Use when starting any conversation or task to determine which planning mode and skill apply. user-invocable: false --- <EXTREMELY-IMPORTANT> If you think there is even a 1% chance a skill might apply to what you are doing, you ABSOLUTELY MUST invoke the skill. IF A SKILL APPLIES TO YOUR TASK, YOU DO NOT HAVE A CHOICE. YOU MUST USE IT. This is not negotiable. This is not optional. You cannot rationalize your way out of this. </EXTREMELY-IMPORTANT> # Planning Workflow You have skills. You MUST use them. Not "should." Not "when convenient." MUST. Invoke relevant skills BEFORE any response or action. Even a 1% chance a skill might apply means you invoke it. If an invoked skill turns out to be wrong for the situation, you don't need to follow it. But you must check. ## The One Rule **When an implementation choice exists, never write code until the human has reviewed and approved a written plan.** Exact mechanical changes may use the exception below. The plan may be inline in the conversation or backed by persisted spec artifacts. Select the mode before invoking an artifact-producing skill, state the choice with one brief reason, and let the human override it. ## Governing Principles Invoking a skill selects relevant guidance. It does not make every pattern in every loaded skill part of the solution. Existing source, explicit requirements, and the smallest correct change determine which patterns apply. Begin with the current design and the minimum behavior-preserving change. Product, infrastructure, and architectural expansion require explicit scope. Keep **ponytail** active throughout development when it is available or required by the repository. Keep **writing-clearly-and-concisely** active whenever writing or editing prose for humans. Before