pydantic/monty · Archived

review-verbosity

Rewrite excessively verbose comments and docstrings added by this branch, and delete tautologous ones. Use after writing a feature to tighten up its comments; this skill edits code rather than only reporting.

First seen Aug 23, 2026

Installation

$ npx skills add pydantic/monty --skill review-verbosity

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Also in this package

Other skills from pydantic/monty.

npx skills add pydantic/monty

Browse all from pydantic/monty

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Repository health

Stars 8.1K
License LICENSE
Default branch main
Open issues 69
Status Archived

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 1,992 B
  • docs SUMMARY.md 232 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 1 installs

SKILL.md

Verbosity review

New comments and docstrings are usually too verbose. Rewrite them concisely. This skill edits — it doesn't just report.

git diff --stat origin/main...HEAD   # which crates changed
git diff origin/main...HEAD

Dispatch one sub-agent per changed area — a crate, or any other group of changed files, so nothing outside crates/ is missed — then review their edits yourself. A single pass over a large diff won't be exhaustive. In every comment the branch touched:

  • Cut it down. Comments and field docstrings rarely over 3 lines, mostly 1; function

and struct docstrings \<= 5.

  • Delete tautology — restating what the code plainly says earns nothing.
  • Delete narration of the obvious, restated type signatures, how the code came to

be, hedging.

  • Keep the motivation — why the code exists and its foot-guns are the valuable part.

Don't cut real information to hit a line count.

  • Drop over-long examples. Public items only, \<= 8 lines, and every one must run in

tests — never ignore.

  • Fix what's out of date. A comment the branch made wrong gets corrected, not left.

Python docstrings are markdown: single backticks, never RST double-backticks.

Only touch what this branch added or changed, unless asked for a wider sweep. Don't change behaviour — if a comment is wrong because the code is wrong, say so rather than rewording it to match. Concision, not terseness: an unfollowable comment is worse than a long one.

make format-rs && make lint-rs
git diff --stat   # format-rs is `cargo +nightly fmt --all` — check it touched only your files

Then report what you cut, by file, in a couple of lines — not a full inventory.