agentleFS
dgalarzadgalarza/agent-skills/skills/conventional-comments/SKILL.md

conventional-comments

Skill72 starsChanged 6 months agoMITView on GitHub

Use this skill when writing, rewriting, or reviewing feedback comments so they follow the Conventional Comments format.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 9 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when feedback needs to be clear, reviewable, and easy to scan.

    line 8
  2. Distinctive:

    Write each comment in this shape:

    line 14
  3. Distinctive:

    Use decorations only when they add useful context.

    line 67
  4. Distinctive:

    Add a decoration only if it changes how the author should prioritize or interpret the comment.

    line 83
  5. Distinctive:

    Write the subject as a concrete, complete sentence.

    line 84
  6. Distinctive:

    Add discussion only when it explains why the comment matters or how to resolve it.

    line 85
  7. Distinctive:

    Avoid vague subjects like "This is wrong" or "Fix this."

    line 93
  8. Very specific:

    Avoid performative praise.

    line 94
  9. Distinctive:

    Prefer one actionable comment over several loosely related concerns.

    line 95
  10. Distinctive:

    Before returning comments, check that:

    line 121

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Conventional Comments

Use this skill when feedback needs to be clear, reviewable, and easy to scan. It is useful for code review, document review, design feedback, RFC comments, editing notes, and any other feedback thread where tone and actionability matter.

Source standard: https://conventionalcomments.org/

Core Format

Write each comment in this shape:

<label> [decorations]: <subject>

[discussion]
  • label is one intent label, such as suggestion, issue, or question.
  • decorations are optional comma-separated context tags in parentheses.
  • subject is the direct comment.
  • discussion is optional supporting context, reasoning, or next step.

Examples:

suggestion: Rename this helper to match the domain term used in the rest of the module.

That makes the call sites easier to connect back to the product copy.
issue (blocking): This branch returns a 200 response when validation fails.

Could we return the existing validation error response here so clients can recover correctly?

Label Selection

Choose the narrowest accurate label:

LabelUse when
praiseCalling out something genuinely good or worth preserving.
nitpickRequesting a trivial preference-based change. Usually non-blocking.
suggestionProposing an improvement. Explain what to change and why.
issueIdentifying a concrete problem. Pair with a suggested fix when possible.
todoFlagging a small necessary task.
questionAsking for clarification because the concern is uncertain.
thoughtSharing a related idea that should not block the work.
choreRequesting a process step required before acceptance. Link the process when possible.
noteHighlighting useful context with no requested action.

Optional expressive labels can be used if they are clearer in context:

LabelUse when
typoFixing spelling or small text mistakes.
polishImproving quality when nothing is strictly wrong.
quibbleCalling out a very small preference without implying importance.

Decorations

Use decorations only when they add useful context.

Common decorations:

DecorationMeaning
(blocking)The work should not be accepted until this is resolved.
(non-blocking)The work can be accepted without resolving this first.
(if-minor)Resolve only if the fix is small or trivial.

Domain decorations can clarify scope, such as (security), (test), (ux), (performance), (docs), or (accessibility). Keep decoration lists short. If the list gets long, split the feedback into separate comments.

Workflow

  1. Identify the review intent before writing the comment.
  2. Choose one label that best matches the intent.
  3. Add a decoration only if it changes how the author should prioritize or interpret the comment.
  4. Write the subject as a concrete, complete sentence.
  5. Add discussion only when it explains why the comment matters or how to resolve it.
  6. For issue, suggestion, chore, and todo, make the expected action explicit.
  7. For uncertain concerns, use question instead of overstating the problem.

Tone Rules

  • Be direct about the requested change.
  • Separate preference from correctness.
  • Avoid vague subjects like "This is wrong" or "Fix this."
  • Avoid performative praise. Use praise only when it is specific and true.
  • Prefer one actionable comment over several loosely related concerns.

Output Patterns

For a single comment, return only the comment:

suggestion (test): Add coverage for the empty input case.

That is the branch most likely to regress because it exits before the normalization step.

For a batch of comments, group them as a list:

- issue (blocking): The migration drops `account_id` before the backfill runs.

  Could we split this into a backfill migration and a later cleanup migration?

- note: This parser now matches the API contract more closely.

For Codex inline code review, use the conventional comment text in the title or body of the code comment, while keeping the comment tied to the smallest useful line range.

Quality Bar

Before returning comments, check that:

  • Each comment starts with <label>: or <label> (<decorations>):.
  • Each label matches the actual intent.
  • Blocking status is explicit when priority could be misunderstood.
  • The subject can stand alone without the discussion.
  • The discussion adds reasoning, context, or a next step.
  • The comment asks for a concrete action when action is expected.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.