How to use the Product Strategy Doc Writer agent skill (ce-strategy)

What is an agent skill, and what does ce-strategy make the agent do?

An agent skill is a folder of instructions and reference files that an AI coding agent loads when a task matches the skill's description. ce-strategy is one such skill, from the EveryInc compound-engineering-plugin project, released under the MIT licence.

When it runs, the agent asks you one question at a time about what the product is, who it serves, how success is measured and where the team is investing. It pushes back on vague answers and keeps the document short. It reads some of the repository, such as the README, docs and recent commits, only to ask sharper questions. The answers come from you, not from the repo.

It can also update an existing STRATEGY.md. When it does, it preserves sections written by others.

• Creates a STRATEGY.md at the repo root when none exists
• Revisits and updates an existing file, for example after a change of direction or roadmap
• Keeps feature plans, schedules and implementation details out of the document

Install the ce-strategy skill

You can install the skill in one of three places. Download the ZIP from the skill's page on this directory and unzip it into a folder your agent reads skills from. The skill's folder is named ce-strategy and contains SKILL.md and a references folder.

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 so it is available to anyone working in that repository.

If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal: ~/.claude/skills/ce-strategy/
• Project: .claude/skills/ce-strategy/ inside your repository
• Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Trigger the skill with a natural request

You do not need a special command. The skill's description tells the agent to use it when you are starting a product, adding a strategy doc, or changing direction or roadmap, so plain requests usually work. You can also pass an optional focus hint, such as 'metrics' or 'approach', to revisit one section.

Ask in your own words, for example that you want to write a strategy doc for a new product, or that your direction has changed and the strategy needs updating.

• "Let's write a STRATEGY.md for this project."
• "We changed direction, can we update the strategy doc?"
• "Revisit the metrics section of our strategy."

What happens during a run

On a first run with no file, the agent first builds a short working model of the product from the repo and shows it to you in chat, asking you to correct it. If the product has no name in the repo, it will ask for one. It then interviews you in a fixed order: purpose, positioning, users, key metrics, tracks, a stress test, boundaries, and optional milestones and brand.

Once every section is captured, the agent drafts the full document, shows it to you, offers one round of edits and then writes STRATEGY.md. If you give a weak answer, expect a follow-up question that quotes your own words back. Two rounds is the most it will push on any one section.

On an update run, the agent summarises the current file in a few lines, lists sections that may be out of date as suggestions, and revisits the one you choose. Other sections are left as they are.

Example: a team starting a new product

A small team with a new repository asks the agent to help write its strategy. The agent reads the README to understand the product, then asks the purpose question. The team answers in their own words, and the agent pushes back on any answer that stays generic. After the interview it shows a draft STRATEGY.md, the team makes one round of edits, and the file is written to the repo root.

Other skills that read the file, such as ce-ideate, ce-brainstorm and ce-plan, will use it as grounding on their next run.

What the skill needs and safety notes

The skill lists no requirements. It does not run scripts and does not use the network, according to its listing. It reads and writes files in your repository, including STRATEGY.md and other project docs it uses for grounding.

Because it edits a file that may be shared, check the draft before accepting it. The skill will not edit a section carrying an author-approved marker, and it reports conflicts or writes a separate linked file instead of editing another person's document.

• Requires: nothing listed
• Runs scripts: no
• Uses network: no
• Writes: STRATEGY.md at the repo root, after you approve the draft

Tips for a useful strategy doc

Keep the document short. The skill pushes back on expansion on purpose, and each section it writes is meant to read in under five minutes together. Put feature plans, schedules and implementation details in a planning or issue tracker tool instead, since the skill keeps them out.

Answer in your own words. The skill quotes your wording back to you, so specific answers produce a sharper document than generic ones.

• Use concrete language about users and metrics rather than broad goals
• Record which metrics matter and where they are tracked, not their current values
• Revisit the strategy after a change of direction so downstream skills read current grounding