Skip to content
Agentmart
Submit a tool

pr-comment-humanizer

Rewrites PR review comments to match the author's terse, direct code review voice.

Skill by edlng

IndexedMIT★ 2
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/edlng-pr-comment-humanizer

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Rewrites PR review comments to match the author's terse, direct code review voice. Strips AI phrasing and produces short imperatives, parenthetical precision, a "nit" prefix for minor things, "we should" framing, verifying questions, and links instead of explanations. Use when drafting or polishing PR review comments before posting.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/pr-comment-humanizer

What's inside

Skill name
pr-comment-humanizer
Description
Rewrites PR review comments to match the author's terse, direct code review voice. Strips AI phrasing and produces short imperatives, parenthetical precision, a "nit" prefix for minor things, "we should" framing, verifying questions, and links instead of explanations. Use when drafting or polishing PR review comments before posting.
License
MIT
Compatibility
claude-code opencode
Instructions
77 lines in SKILL.md · view on GitHub

Agent Skills spec issues

  • allowed-tools: must be a space-separated string

Files (1)

  • SKILL.md

SKILL.md (first lines)

# PR Comment Humanizer

Rewrite PR review comment text to match this specific voice: terse, direct, imperative. No warmup, no hedging, no AI vocabulary.

## Voice Profile

Derived from real PR comments at https://github.com/edlng/ChatDev/pull/1.

**Core traits:**

- **Brutally terse.** If one line does it, use one line. "valkey-glide 2.4" not "Please update the dependency version to valkey-glide 2.4 to ensure compatibility."
- **Imperative by default.** "Add another note about X" not "It would be good to add a note about X" or "Consider adding a note about X."
- **"Might as well"** for low-effort improvements worth doing while you're there. "Might as well also add a case for an upper bound."
- **"nit"** prefix (lowercase) for minor stuff that won't block merge. Never "minor nit" or "small nit", just "nit".
- **"Same concern as above"** to cross-reference instead of repeating yourself.
- **Parenthetical precision.** Use `(i.e., ...)` and `(e.g., ...)` inline rather than new sentences.
- **"We should"** not "you should". It treats the problem as shared.
- **Questions to verify, not assert.** "Can you confirm that X?" not "X is incorrect."
- **Links instead of explanations.** Drop the doc URL and stop. Don't summarize what's at the link.
- **Backtick code references inline.** Wrap identifiers, values, and branch names in backticks.
- **No period on single-line comments.** Multi-sentence comments get periods; one-liners often don't.

## What to strip

Apply the `unslop` skill's writing rules first. Then apply the voice rules in this skill, including these PR-comment-specific ones:

- Do not use "consider". Use the imperative or "might as well".
- Do not use "it would be good to". Say what to do.
- Do not use "please". It's a code review, not a polite request.
- Do not use "this looks good but". Lead with the finding.
- Do not use "I noticed that" or "I see that". State the observation.
- Cut trailing "to ensure X" clauses. The reader knows why.
- Do not use "as per". Use "per" or rewrite.
- Leave severity labels (Blocking/Recommended/Nit) out of the body. Word choice and the "nit" prefix carry severity.

## Process

1. Read the comment body.
2. Apply the `unslop` writing rules.
3. Strip the PR-comment-specific patterns above.
4. Rewrite using the voice profile.
5. Self-check: read it aloud. Would a real engineer write this in a 2am code review?
6. If not, cut more.

## Examples

**Before:**
> I noticed that you might want to consider adding validation for the upper bound as well (i.e., values greater than 65535). This would ensure that the error message is more general and covers all invalid port scenarios. It would be good to change the error to something like "must be a valid port 1-65535" to make it clearer for users.

**After:**
> Might as well also add a case for an upper bound (i.e., `> 65535`) and change the error to be more general (e.g., must be a valid port 1-65535).

---

**Before:**
> This is a great start! However, I noticed that the tests for the two scenarios are in separate test methods. It would be better to consolidate these into a single test to improve readability and reduce redundancy.

**After:**
> Collapse this into one test

---

**Before:**
> Please note that this PR appears to be targeting the `main` branch. We should follow proper git practices and merge into the `add-valkey` branch instead, especially since we plan to open a PR from `edlng:add-valkey` to `upstream:main` later on. You may need to reopen another PR to change the target branch.

**After:**
> We should be merging from your branch into `add-valkey` instead of into `main`. It's proper practice especially once we open a PR on the main repo (`edlng:add-valkey` -> `upstream:main`). You may have to reopen another PR.

---

**Before:**
> I wanted to flag that the same issue exists here as in the previous comment regarding the database index validation. Valkey supports 16 logical databases numbered 0-15. You can find more information about this in the official Valkey documentation.

**After:**
> Same concern as above comment. Valkey has 16 logical databases 0-15 (https://valkey.io/commands/select/).

From edlng/agents, MIT.

Signals

Stars
2
Forks
0
Open issues
0
Last commit (this folder)
Oct 2, 2026 (27 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from edlng

Cite this page

pr-comment-humanizer skill. Agent Mart. https://agentmart.dev/skills/edlng-pr-comment-humanizer (last updated 2026-10-24)