Skip to content

modem-dev/ossrules / public/files/astro/.agents/skills/writing-comments

writing-comments

Unverified

How to write JSDoc (/** */) and inline (//) comments in the Astro codebase, for contributors reading the source — not end users. Use whenever writing or editing comments in .ts/.js source, including comments added incidentally while fixing bugs or building features. Does not cover the @docs-generated config/error reference.

Writing Comments

Purpose

Comments in this repository are read by contributors, months or years after they were written, with none of the context you have right now. This skill defines who that reader is, what each kind of comment is for, and which patterns are banned.

Scope Boundary

This skill governs contributor-facing comments in the TypeScript/JavaScript source. It does not apply to end-user documentation:

Everything below is about the source a contributor reads at HEAD.

The Reader

Write for an Astro contributor who is competent in TypeScript but has no access to your current context: not this conversation, not the pull request, not the issue, not the diff. They see only the repository at HEAD.

Two consequences follow directly:

  1. Never narrate change history. Words like "now", "previously", "no longer", "the new approach" are meaningless at HEAD, where only one approach exists. State how the code works, not how it came to be. (A @deprecated notice is the exception — see Conventions — because it describes the contract's future, which the reader needs.)
  2. Never address the reviewer. A comment that argues your change is correct ("this properly handles X") belongs in the PR description, not in the source. The comment must justify the code as it stands, permanently.

Three Kinds of Comments, Three Different Jobs

KindSyntaxJobContains
File / module overview/** */ at the top of the fileExplanationWhy the module exists, the concepts and terms it defines, how the pieces relate, design rationale
Item docs/** */ directly above a declarationReferenceThe contract: behavior, parameters, return value, thrown errors, invariants. Neutral and factual
Inline comments// inside a bodyRationaleOnly what the code cannot say: constraints, workarounds (with issue links), non-obvious coupling

Do not mix the jobs. Implementation details do not belong in the /** */ contract — put them as // comments inside the body. The contract does not belong scattered across inline comments — put it on the declaration.

The Deletion Test

Before writing any comment, ask: does this state something the reader cannot recover from the code itself?

  • If the information is already carried by names, types, or structure, do not write the comment. If the name fails to carry it, improve the name.
  • Information that legitimately needs a comment: an invariant, a rationale, a coupling to code elsewhere, a workaround with a link, surprising behavior of a dependency, a term of art the module defines.

When editing later, the same test applies in reverse: a comment that no longer passes it should be deleted, not left to rot.

Link to the Issue for Workarounds

This codebase consistently anchors workarounds to a source. Any comment that explains a workaround, a HACK, a regression guard, or surprising behavior of a dependency must link the GitHub issue or PR that motivates it. The link is what lets a future reader tell whether the workaround is still needed.

// Handle recommended nanostores. Only @nanostores/preact is required from our testing!
// Full explanation and related bug report: https://github.com/withastro/astro/pull/3667
'@nanostores/preact',

A workaround with no link is indistinguishable from a mistake.

Banned Patterns

Narrating the next line. Delete these on sight:

// Increment the generation counter
generation += 1;

Change-history narration. Rewrite as present-tense rationale:

// BAD: We now resolve lightningcss from the user's root instead of ours.
// GOOD: lightningcss is an optional peer dep, so it resolves from the user's project root.

Reviewer-addressed justification. Move the argument to the PR:

// BAD: This correctly handles the multi-encoded path from the bug report.
// GOOD: A path still encoded after MAX_DECODE_ITERATIONS is rejected, so
//       middleware and routing can never disagree on the decoded path.

Restated JSDoc. A /** */ block that rewords the declaration name says nothing:

// BAD:
/** Compiles the styles. */
function compileStyles(...)

// GOOD:
/** Rewrites relative `url()` references in `css` against `base`, leaving
 *  absolute and data URLs untouched. */
function compileStyles(...)

Vague hedging. "Some cases", "various reasons", "handles edge cases", "etc." — either name them or drop the sentence.

Emojis. Banned in source, comments included (repo-wide policy).

Ad-hoc section banners (// ----- helpers -----, // ==== TYPES ====). This codebase has no // #region folding convention; do not add banners. If a file is long enough that you reach for one, that is a signal to split the file, not to decorate it.

Conventions in This Codebase

JSDoc tags. @param name - description, @returns, and @throws state the contract. Brace-wrap a type (@returns {Promise<string>}) only when the signature alone is ambiguous. Use @example with a fenced ```js block for non-obvious usage.

Cross-references. Use {@link Symbol} / {@linkcode Symbol} rather than a bare name, so a rename updates the reference and editors can jump to the target.

@internal. Marks API that is not part of the public surface. It is a convention only — there is no typedoc or api-extractor here to strip it — so it documents intent but does not replace access modifiers.

@deprecated. State the migration, then the removal horizon:

/** @deprecated Use the instance method `cookies.consume()` instead. This will be removed in Astro 7 */

Say what to use instead, not merely that the symbol is deprecated. This future-tense note is contract information the reader needs, not banned change-history narration.

TODO. Use // TODO: for deferred work; link an issue when one tracks it. For work gated on a breaking-change window, use the established idiom // TODO: remove in Astro <N>. There is no FIXME in this codebase — do not introduce it.

Editing Existing Code

  • Preserve existing comments. If your change alters behavior, extend or correct the specific prose — never replace it with generic text. Deleting hard-won context is worse than leaving a comment slightly stale.
  • When your change makes a comment false, fix it in the same diff. A stale comment is worse than none.
  • Match the surrounding density. A heavily documented module deserves the same level on new items; do not blanket a sparse module with comments.

Self-Check Before Finishing

After completing any task that touched comments, re-read only the comments in your diff, in isolation from the code changes:

  1. Does each one pass the deletion test?
  2. Does any reference the conversation, the change itself, or the reviewer?
  3. Does every workaround link its issue or PR?
  4. Would a reader without access to the diff understand each one?

Fix or delete what fails. Deletion is the default; a missing comment is cheaper than a misleading one.

References

rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 464kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 196kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 196kUpdated Claude CodeCodexVerifiedInstall