agentleFS
foryouself83foryouself83/harness-tier/skills/prose-review/SKILL.md

prose-review

Skill10 starsChanged 32 days agoApache-2.0View on GitHub

Use when comments, docstrings or documents need checking against the prose rules — before writing them, or to clean up what is already there. The doc-sync gate calls it.

2 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

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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
  1. Very specific:

    Keep this resolved list; step 4 reuses it.

    line 36
  2. Distinctive:

    No pattern answers these. Read each comment, docstring and paragraph the paths carry:

    line 48
  3. Distinctive:

    Run this over the same list step 1 resolved, not a freshly re-derived changed-file list — the fix itself can create or rename files, and …

    line 65

≈ “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.

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

prose-review

Half of the prose rules are patterns and half are judgement. doc-style.md is the rule; this skill runs both halves over files you name.

Report every violation in the language the user is writing in, and write what follows each box key the same way. The three box keys — CRITICAL TRAP:, Trigger:, Symptom: — stay literal in every project: the checker parses them. The rule file is English; the reader may not be.

1. The machine half

$ARGUMENTS names the paths, space-separated. Empty means the caller gave none — take the changed files, narrowed to the project's doc-style scope (its paths and exclude, or every .md, .py and .sh where doc_style is off or absent), so nothing the project excluded is rewritten. Run it from the repository root; it prints root-relative paths, one per line:

{ git diff -z --name-only HEAD; git ls-files -z --others --exclude-standard; } |
  xargs -0 python3 .claude/harness-tier/scripts/doc_style_check.py --scope

The lists travel NUL-separated: a path holding a space stays one argument.

An unrecognized arguments error means the host copy of the script predates --scope, and a repo without the script has nothing to run: say so, tell the user to re-run /flow-init, and take the two git lists unnarrowed. Never continue with an empty list the failure left.

Keep this resolved list; step 4 reuses it. Quote each path wherever a command below takes <paths>.

python3 .claude/harness-tier/scripts/doc_style_check.py --lint <paths>

ANCHOR · META · TRAP · HIST · SHA · PLAN · FILLER · ENDING are decided here, and a repo without the script skips this step and keeps the judgement half.

2. The judgement half

No pattern answers these. Read each comment, docstring and paragraph the paths carry:

  • Does it explain how? A paraphrase of the code beneath it goes, whatever it costs.
  • Could the code raise instead? A constraint stated in prose that an assert, a validated bound or a type could carry is a comment that should not exist. Propose the check, not a better sentence.
  • Is it self-evident? A comment restating its own identifier is noise.
  • Is a number a measurement or a contract? Anything else is a guess wearing a figure — give the direction instead.
  • Is it as short as it can be? Nominal endings, no connective padding.

3. Fix

Show each violation with the replacement, grouped by file. Apply what the user accepts.

4. Prove nothing was lost

Run this over the same list step 1 resolved, not a freshly re-derived changed-file list — the fix itself can create or rename files, and re-deriving here would verify a different set than the one already rewritten. Skip this step when that list is empty: --verify-git errors on no arguments, where --lint above silently no-ops on the same case.

python3 .claude/harness-tier/scripts/doc_style_check.py --verify-git <paths>

A rewrite that drops a heading, a fenced block, a URL or an inline-code span is a rewrite that lost the fact. Restore it, or say in the report why the removal was the point.

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.