Guide · Agent skill
How to use the Structured Planning agent skill (ce-plan)
Updated · written with AI from the project's own documentation and checked against our catalogue data
ce-plan is an agent skill from the EveryInc compound-engineering-plugin that makes a host AI agent write a structured plan before it starts a multi-step task. It covers software work and non-software work, and it is designed to plan rather than implement. After this guide you will know what the skill does, how to install it from the ZIP, how to trigger it with plain requests, and what it needs and touches when it runs.
Structured PlanningEveryInc/compound-engineering-plugin · MIT Download the skill Quick start
- 1Download the ZIP. Download the ce-plan ZIP from this directory page.
- 2Unzip the skill. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
- 3Or upload it in the Claude apps. Go to Settings → Capabilities → Skills and upload the skill there.
- 4Check the references folder. Confirm the references folder and SKILL.md are present, since the skill stops if its reference files cannot be read.
- 5Make a planning request. Ask in plain language, for example to plan an implementation or break down a requirements doc.
- 6Review the plan. Read the saved plan and check the steps, order, checks and constraints before building.
- 7Build with ce-work. Once you approve the plan, move to the separate ce-work step to carry it out.
What the ce-plan skill makes the agent do
Without a plan, an agent often starts changing things as soon as it gets a messy multi-step request. The ce-plan skill changes that order. Once it is installed, the agent first writes a structured plan that sets out what to do, in what order, how to check each step, and which constraints must be kept.
The skill researches the codebase or the requirements before it chooses an approach, so decisions are based on evidence rather than guesses. It is built to plan only. Building the work is a separate step, handled by the ce-work skill.
- Turns a feature request or requirements document into ordered steps.
- Lists how to verify each step and which constraints must be preserved.
- Works for software tasks and for non-software projects.
- Can deepen an existing plan instead of writing a new one.
Install the ce-plan skill from the ZIP
Download the ZIP for the ce-plan skill from this directory page, then unzip it. The skill folder contains a SKILL.md file and a references folder. Both must be present.
For a personal install that applies to all your projects, unzip it into ~/.claude/skills/. For a single project, unzip it into that project's .claude/skills/ folder.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload
Trigger the skill with natural requests
You do not need a special command. Ask the agent in ordinary language and the skill's description tells the agent when to use it. The description names these triggers: asking to plan, break down implementation, plan from requirements, or deepen an existing plan.
If you want to explore an idea before deciding what to build, the skill description says to prefer ce-brainstorm for exploratory framing. Use ce-plan once you know roughly what you want built.
- "Plan the implementation of the CSV export feature."
- "Break down this requirements doc into steps: docs/requirements/billing.md."
- "Deepen the existing plan in this file."
- "Plan the steps to reorganise our volunteer onboarding process."
What the skill needs to run
The skill needs the host agent's file-writing ability, because the plan is saved as a document. It can also use optional subagent dispatch. The reference files in the references folder must be present, including files such as output-mode, resume, intake, research and structure. If the skill cannot read them, it stops.
Two helpers are optional. The slack-researcher helper needs a Slack connection, and the web-researcher helper needs web access. You only need these if you use those helpers.
- File-writing access in the host agent.
- Optional subagent dispatch.
- All files in the references folder, including references/agents/ and SKILL.md.
- Optional: a Slack connection for slack-researcher.
- Optional: web access for web-researcher.
Safety notes: scripts and network access
The skill does not run scripts. It does use the network, which is why the optional helpers exist. Its own rules say it must not write production code, run tests, or implement the work while planning.
Check the plan before you act on it. The skill is not a substitute for expert review on high-risk areas. Its own guidance flags authentication, payments, migrations and external contracts as risk surfaces that need careful checking.
- Runs scripts: no.
- Uses network: yes, for research helpers.
- Does not implement code while planning.
- Plans on authentication, payments and migrations need expert review.
Tips for better plans
Give the agent the source material. If you have a requirements document, name its path so the plan builds from it. If a plan already exists, name that file and ask the agent to deepen it.
Use ce-plan for work that has several steps or a decision to weigh. For one-line changes, asking directly is usually faster.
- Name the requirements doc or plan file path in your request.
- Say whether the work is software or non-software.
- Review the plan before asking the agent to start building with ce-work.
- Skip the skill for quick one-line changes.
What people use it for
- Breaking a feature request into ordered implementation steps before any code is written.
- Turning an existing requirements document into a checkable implementation plan.
- Deepening a plan that already exists by adding research and detail.
- Planning a non-software project, such as a process change, as a clear sequence of checkable steps.
Frequently asked questions
How do I install the ce-plan skill?
Download the ZIP from this directory page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.
Is ce-plan free to use?
The skill is published under the MIT licence. This page does not list pricing, so check the licence and the repository for terms before relying on it.
What does ce-plan require to run?
It needs the host agent's file-writing access and the reference files in the references folder, including output-mode, resume, intake and research. Subagent dispatch is optional, and Slack or web access is only needed if you use the slack-researcher or web-researcher helpers.
Does ce-plan write code?
No. The skill is designed to plan only. It researches and writes the plan, and the building happens in the separate ce-work step.
What is the difference between ce-plan and ce-brainstorm?
The skill description says to prefer ce-brainstorm for exploratory framing. Use ce-brainstorm to work out what to build, and ce-plan to plan how to build it.
Why does the skill stop with an error about missing files?
The skill stops if it cannot read its reference files. Check that the references folder and SKILL.md were unzipped alongside each other inside the skill folder, then try again.
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
ce-plan suits developers who want an implementation plan reviewed before code is written, teams turning a requirements doc into steps, and people planning non-software projects who want a checkable sequence. Skip it for quick one-line changes, and do not rely on its output alone for authentication, payments or migrations without expert checking.
Download the skill