Guide · Agent skill
How to use the Collaboration Comment Writer agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The comment-writer skill is an agent skill that tells an AI agent how to write comments that a person will read, such as pull request feedback, issue replies, review notes and Slack or Discord updates. It is for engineers, maintainers and team leads who want drafts that are short, friendly and direct. After this guide you will know what the skill changes, how to install it, how to ask for it in plain language and what it does and does not do.
Quick start
- 1Download the ZIP. Get the comment-writer skill ZIP from the item page on this directory.
- 2Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use, or into the project's .claude/skills/ folder for one project.
- 3Or upload in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the skill.
- 4Ask in plain language. Describe the comment you need, the platform (PR, issue, Slack) and the outcome you want.
- 5Review and post. Read the draft, check any technical claims yourself, then post it.
What does the comment-writer skill make the agent do?
Once the skill is loaded, the agent follows a fixed set of writing rules whenever it drafts a comment for a human to read. The rules ask it to start with the actionable point, not a recap of the whole pull request, and to explain the technical reason briefly when requesting a change.
The skill also asks the agent to keep comments to one to three short paragraphs or a tight bullet list, to avoid pile-ons by commenting on the highest-value issue rather than every small preference, and to avoid em dashes. It tells the agent to write in the language of the surrounding thread, and to use neutral, professional Spanish by default for Spanish threads unless the user or context clearly calls for a regional tone.
- Leads with the actionable point
- Gives the technical reason when asking for a change
- Keeps messages short, in a few paragraphs or bullets
- Matches the language of the thread
- Avoids em dashes
How do you install the comment-writer skill?
Download the skill ZIP from the item page on this directory. The skill is a single file named SKILL.md. Unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder if you want the skill to apply only to that project.
If you use the Claude apps instead, you can upload the skill there under Settings, then Capabilities, then Skills.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills, then upload
How do you trigger the skill with a natural request?
You do not need a special command. The skill describes its own triggers: PR feedback, issue replies, reviews, Slack messages and GitHub comments. Ask for the kind of comment you need in ordinary words and mention the context so the agent can match the language and tone.
The skill is meant for comments another human will read. It is not designed for code decisions, so describe the change you want commented on and state the outcome you want, such as approval, a request for changes or a split.
- Draft a review comment asking for this PR to be split into two commits
- Write a reply to this GitHub issue that thanks the reporter and asks for a reproduction
- Write a Slack update on the migration status for the team channel
- Reply to this Spanish issue thread in Spanish
What examples does the skill produce?
The skill includes three example patterns. A requested change explains the problem and why it matters, then names the next step. An approval gives a clear verdict and a short follow-up note. A request to split a large change names the size limit and suggests an order for the pieces.
The examples in the skill follow the same three-part formula: a direct observation or request, the reason it matters if needed, and a concrete next action.
What does the skill need, and is it safe to use?
The skill includes one suggested command that inspects a pull request with the GitHub CLI (gh) before you write review feedback. That command needs gh installed. The rest of the skill is written instructions only. According to this listing, the skill runs no scripts and uses no network access.
Because the skill only shapes writing, check every draft before posting it. Its style rules do not check whether your technical judgement is correct, and the listing says the skill is not suited to formal legal, HR or customer-facing messages that need institutional review.
- GitHub CLI (gh) is needed only for the optional PR inspection command
- Runs scripts: no
- Uses network: no
- Review drafts before posting, especially for technical claims
gh pr view <PR_NUMBER> --json title,body,additions,deletions,changedFilesWhat people use it for
- Drafting a pull request review that asks for a change and explains the reason in two sentences
- Replying to a GitHub issue in a calm, non-robotic tone as a maintainer
- Writing a short async project update for a Slack or Discord channel
- Asking an author to split an oversized pull request into reviewable pieces
Frequently asked questions
Is the comment-writer skill free to use?
The item page lists the licence as MIT, but the SKILL.md file in this listing declares Apache-2.0 in its metadata. Check the licence file in the source repository before relying on either one for your project.
What do I need to install the comment-writer skill?
You need the SKILL.md file from the downloaded ZIP and a Claude setup that loads skills, either a personal or project skills folder or the Settings, Capabilities, Skills upload in the Claude apps. The GitHub CLI is only needed for the optional pull request inspection command.
Where do I put the comment-writer skill folder?
Unzip it into ~/.claude/skills/ to make it available in your personal setup, or into a project's .claude/skills/ folder to limit it to that project.
Does the comment-writer skill review my code?
No. The listing says the skill governs tone and structure, not correctness. It will not decide whether a change is technically right, so check the technical content yourself.
Does the skill run scripts or access the internet?
According to this listing, it runs no scripts and uses no network access. Its one suggested command, gh pr view, is something you run through the GitHub CLI.
What are the alternatives to this skill?
You can write comments without a skill, or give your agent your own style guide in a prompt. This listing does not compare the skill with other tools, so judge the fit by the rules it sets out.
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 this skill if you write many pull request reviews, issue replies or team updates and want short, friendly, consistent drafts. Pick something else for formal legal, HR or customer-facing messages, or whenever you need the agent to judge whether a technical change is correct.
Download the skill