Guide · Agent skill
How to use the Brainstorm a Feature agent skill (ce-brainstorm)
Updated · written with AI from the project's own documentation and checked against our catalogue data
ce-brainstorm is an agent skill for Claude Code that helps you work out what to build before any code is written. It is for developers and product-minded engineers who have a fuzzy feature idea and want it turned into clear requirements, scope boundaries and success criteria. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what it needs and where it stops.
Brainstorm a FeatureEveryInc/compound-engineering-plugin · MIT Download the skill Quick start
- 1Download the ZIP. Download the ce-brainstorm ZIP from this page.
- 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ inside a project to limit it to that project.
- 3Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.
- 4Open a Git repository. Work inside a Git repository, since the skill resolves paths with git rev-parse.
- 5Describe the idea. Tell the agent what you want to explore and ask it to brainstorm or scope the feature.
- 6Answer the questions. Reply to one focused question at a time until the scope and success criteria are clear.
- 7Take the output. Use the chat summary for small work, or the plan file under your docs root for larger work, and hand it to ce-plan if you need technical detail.
What the Brainstorm a Feature skill makes the agent do
When you ask the agent to brainstorm or scope something you want to build, the skill runs a structured dialogue. It asks focused questions, sizes the work, and records what to build: requirements, scope boundaries and success criteria. It does not implement code.
The output depends on the size of the request. Small, well-bounded requests end as a short chat summary with no file. Larger requests produce a requirements-only unified plan under a plans folder inside your docs root. A later planning step, ce-plan, can extend that same document with technical details.
- Asks one question per turn, and only about decisions the environment cannot settle
- Sizes the work before deciding whether a file is needed
- Records requirements, scope boundaries and success criteria
- Hands off to ce-plan for the how, or to other Compound Engineering skills such as lfg
How to install the ce-brainstorm skill
Download the ZIP for this skill from this page. Then install it in one of two ways. For a personal install, unzip it into ~/.claude/skills/. To limit it to one project, unzip it into the .claude/skills/ folder of that project.
If you use the Claude apps instead, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload
How to trigger the skill with natural requests
You do not need a special command. Describe what you want to explore and say that you want to brainstorm or scope it. The skill matches requests where the user wants to brainstorm or scope what to build.
The skill is not for work that is already specified. If you ask it to implement a clearly defined change, it is designed to stop and route you to an implementation workflow instead of building.
- 'Let's brainstorm a way to let users export their dashboard data.'
- 'I have a vague idea for notifications. Help me scope what we should build.'
- 'Scope the requirements for a bulk-edit feature before we plan it.'
- Add the optional argument [output:html] if you want the plan written as HTML instead of markdown.
What the output looks like and where it goes
A plan file is written only when the dialogue produces a decision that a later reader needs recorded under a stable ID, or when you ask for one. Plans go in a plans folder under your docs root, which defaults to docs if you do not set a value. The file name includes the date, the time, and the topic.
The format is markdown or HTML, never both. The plan contains a Goal Capsule and the Product Contract. It does not contain a Goal Launch Block or Reader Index.
- Default docs root: docs, unless set in .compound-engineering/config.yaml
- Set docs_root in .compound-engineering/config.yaml to change it
- Lightweight work ends in a chat paragraph with no file
<root>/plans/YYYY-MM-DD-HHMM-<type>-<topic>-plan.<md|html>What the skill needs and what it does not do
The skill needs to run inside a Git repository, because it resolves paths with git rev-parse. A config file at .compound-engineering/config.yaml is optional and is used for a custom docs_root and output settings.
Companion Compound Engineering skills such as ce-plan, lfg and ce-pov are used for handoffs. The skill runs no scripts and uses no network.
- Requires: a Git repository
- Optional: .compound-engineering/config.yaml
- Companion skills for handoffs: ce-plan, lfg, ce-pov
- Runs scripts: no
- Uses network: no
Safety notes and when to pick a different skill
Because the skill runs no scripts and uses no network, its effects are limited to dialogue and to writing plan files inside your docs root. It does not change your code. Review any plan file it writes before you share it.
Two kinds of request are routed elsewhere. Non-software brainstorming is handled by a different reference file. A verdict on adopting a named external technology is handed off to the separate ce-pov skill.
- Non-software brainstorming: routed to a different reference, not handled here
- Verdict on a named external technology: use ce-pov
- Already-specified implementation work: use an implementation workflow
Tips for better brainstorming sessions
Give the skill the problem, not only the solution. Answer its questions in order, since it asks one at a time. If you want outside context from your team, the Slack researcher is opt-in and must be requested.
Keep the request focused. A narrow idea produces a short chat summary, while a broad one may produce a plan file that a later planning step builds on.
- State the user problem before the feature idea
- Answer one question at a time to keep the dialogue clear
- Ask for a plan file explicitly if you want one
- Run ce-plan afterwards to add technical detail
What people use it for
- Clarifying a fuzzy feature idea before anyone writes code
- Producing written requirements that a planner or reviewer can use
- Setting scope boundaries and success criteria for a feature in a team that runs spec-before-build
- Deciding whether a request is small enough for a chat summary or needs a plan file
Frequently asked questions
How do I install the ce-brainstorm skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for a personal install, or into .claude/skills/ inside a project. In the Claude apps, you can instead upload it under Settings, Capabilities, Skills.
Is the ce-brainstorm skill free to use?
The skill is published under the MIT licence, as listed on this page. Check the repository licence file for the full terms before you reuse or redistribute it.
What does ce-brainstorm require to run?
It requires a Git repository, because it resolves paths with git rev-parse. A .compound-engineering/config.yaml file is optional, and companion skills such as ce-plan, lfg and ce-pov are used for handoffs.
Does ce-brainstorm write code?
No. The skill records requirements, scope boundaries and success criteria and does not implement code. Technical detail is added later by ce-plan.
What is the difference between ce-brainstorm and ce-plan?
ce-brainstorm decides what to build through dialogue, while ce-plan adds how to build it to the same unified plan. Use brainstorm first when the idea is still fuzzy.
Why did the skill tell me to use ce-pov instead?
ce-pov handles verdicts on adopting a named external technology. ce-brainstorm hands those questions off to it rather than answering them.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use ce-brainstorm if you have a fuzzy feature idea and want it turned into clear, written requirements before coding, especially if you work in a Git repository with Claude Code. Pick a different tool if the work is already specified, if you need non-software brainstorming, or if you want a verdict on adopting a specific external technology.
Download the skill