add-comment
Skill1 starsChanged 3 months agoMITView on GitHub
Add high-signal comments to existing or just-completed code changes. Comments only, never change behavior. Judge the code's purpose first, then pick a level — L1 concise (engineering default), L2 detailed (critical paths / handoff), L3 teaching (learning code, explain every symbol). Comment language follows the target codebase's existing comment language — never default to Chinese or English. Use when the user asks to "add comments", "comment this", "explain every X", or "make it readable like prose".
1 other project uses a version of this.
Adopted by
Most recently taken up by
- alexchen5/research-epic-managertheir own version, first seen Aug 31, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- alexchen5/research-epic-manager★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository28 other files it gives its agents.
- agent-supervisorSkill
- ai-seoSkill
- anti-goalsSkill
- bestwaySkill
- codebase-renderSkill
- competitor-profilingSkill
- copywritingSkill
- decision-reviewSkill
- directory-submissionsSkill
- explain-and-saveSkill
- happy-featSkill
- i-have-adhdSkill
- investment-researchSkill
- launchSkill
- learning-new-languageSkill
- lock-decisionsSkill
- moleSkill
- onboardingSkill
- open-source-reviewSkill
- pmSkill
- product-marketingSkill
- product-startupSkill
- prototype-treeSkill
- read-sessionSkill
- reverse-engineer-marketingSkill
- socialSkill
- understandSkill
- update-harnessSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
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
- Distinctive: line 8
Add high-signal comments so future readers quickly grasp *why it's written this way, where data comes from, how state flows, how to verify* …
- Distinctive: line 15
Never default to Chinese or English.
- Distinctive: line 38
Never restate what the code already shows.
- Very specific: line 58
Default reader profile: frontend (JS/TS) background unless stated.
- Distinctive: line 61
Verify facts (API semantics, defaults, units) against source/docs first.
- Distinctive: line 77
Treat TODOs as facts; unverified contracts get "unverified" or nothing.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
add-comment
Add high-signal comments so future readers quickly grasp why it's written this way, where data comes from, how state flows, how to verify — not a prose translation of the code.
Comment language
Match the dominant language of existing comments in the target file/repo. Never default to Chinese or English. If the codebase has no comments, follow its docs/README language; if still ambiguous, follow the user's request language and say so in the report.
Levels
Judge the purpose first, pick a level, state the choice in the report. The user's explicit level always wins.
| Signal | Level |
learn/tutorial/playground dirs, examples/ exercises, study-note code | L3 |
| "explain every X", "make it readable like prose", "teach me" | L3 |
| production/business code, daily bugfixes | L1 |
| cross-module paths, handoff, complex state machines, async orchestration | L2 |
Mixed cases: judge per file. Unjudgeable → L1, with the reasoning stated.
L1 concise (engineering default)
- Comment only the non-obvious: boundaries, data contracts, persistence, async side effects, fallbacks, ordering dependencies.
- 2–3 lines each: why this way, where the data comes from.
- Never restate what the code already shows.
// Session restore replays only server-persisted data; component-local
// state is lost. Selection must go into the persisted payload.
const saved = data.transferState;L2 detailed (critical paths / handoff)
L1 plus: entry responsibilities and callers; field provenance across adapters/props; state ownership (who writes / who reads / when it changes); timing (sync vs async, fire-and-forget vs awaited, rollback on failure). The reader is assumed fluent in the language — no syntax teaching.
L3 teaching (learning code)
L2 plus symbol level: every syntax/keyword/type/API the target reader may not know.
- Default reader profile: frontend (JS/TS) background unless stated; map to
known concepts only when accurate (
fold ≈ reduce,Option ≈ typed null). - File-header overview; section banners; per-function purpose; branch rationale.
- Verify facts (API semantics, defaults, units) against source/docs first.
- High density allowed; two bottom lines hold: never change behavior, never invent facts.
// Unlike JS, `let` bindings are immutable by default;
// `let mut` is required to mutate.
let s = String::from("hello");Do not
- Change logic, names, control flow, or formatting while commenting.
- Invent interfaces / fields / config keys.
- Zero-information comments ("set the variable", "call the function") — except L3 syntax explanations, which are reader knowledge, not restatement.
- Treat TODOs as facts; unverified contracts get "unverified" or nothing.
- Rewrite production files at L3 density unless explicitly asked.
Verify
git diff --check -- <files>; run lint/build when comments could affect
syntax (JSX/TSX); cargo check or equivalent for compiled languages.
Report
Purpose judgment and level choice; files touched; comments-only confirmation; verification run; implementation risks the comments surfaced but didn't fix.
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
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.

