Skip to content
Agentmart
Submit a tool

github-pr-comments

Draft, refine, place, or post actionable GitHub pull-request review comments from validated findings

Skill by OrestesK

IndexedMIT
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/orestesk-github-pr-comments

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

Draft, refine, place, or post actionable GitHub pull-request review comments from validated findings

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/github-pr-comments

What's inside

Skill name
github-pr-comments
Description
Draft, refine, place, or post actionable GitHub pull-request review comments from validated findings
License
MIT
Instructions
174 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# GitHub PR Comments

Turn validated review findings into concise, useful GitHub comments.

This skill owns comment selection, wording, placement, severity, review event, approval, and posting verification. It does not perform the underlying code review.

## Preflight

Before drafting comments:

- Identify the repository, pull request, and current head revision
- Confirm that each finding and its evidence apply to the current head
- Read existing comments to avoid duplicate or noisy feedback
- If the head changed after review, return affected findings for revalidation
- Do not silently weaken a finding when its original evidence is unavailable

## Finding Filter

Keep only findings that pass every applicable gate:

- **Evidence:** direct source, diff, test, documentation, or check evidence supports the claim
- **Impact:** the comment explains why the issue matters
- **Actionability:** the author can fix it, answer a focused question, or intentionally decline it
- **Scope:** the issue belongs to this pull request
- **Signal:** the value exceeds the notification and review-noise cost
- **No speculation:** the claim does not depend on an invented or unrealistic scenario
- **Assumptions:** every material assumption is explicit

When a claim depends on an uncertain assumption, ask a focused question instead of stating the assumption as fact. Drop it when the assumption is weak, unverifiable, or not worth asking about.

Drop or downgrade:

- Linter or formatting issues automation should catch
- Personal taste without an established repository pattern
- Cosmetic or minor nits without concrete benefit
- Large refactors not required for the current pull request
- Generic architecture essays
- Findings unrelated to changed behavior
- Claims that cannot be verified
- Several comments describing the same root cause

Prioritize correctness, simplification, architecture, ownership, deduplication, typing boundaries, and reuse of established patterns.

## Choose Placement

Use an inline comment for:

- A concrete issue on a changed line
- A local simplification at one location
- A type or API problem visible at a specific declaration
- A test or documentation issue anchored to a changed line

Use the review body or a general pull-request comment for:

- Cross-cutting architecture feedback
- Findings spanning multiple files
- Scope or product questions
- A valid finding with no commentable changed-line anchor
- A concise review summary

Do not force an inline comment onto an unrelated or invalid anchor.

## Write the Comment

Keep each comment concise but complete.

For a nontrivial finding, use the parts that add information:

```markdown
<focused statement or question>

What’s wrong:
- <specific issue>

Evidence:
- <file:line or direct evidence>

Impact:
- <observable consequence or maintenance cost>

Assumption:
- <only when the claim depends on one>

Suggested fix:
- <smallest coherent direction>
```

For an assumption-dependent or product decision, replace `Suggested fix` with:

```markdown
Question:
- Assuming <specific assumption>, should we <recommended direction>?
```

Tiny comments may use one short paragraph. Do not force every heading into every comment.

## Language

Use casual, direct, code-focused language.

Prefer:

- “Can we move this into the existing helper?”
- “This duplicates the ownership already in …”
- “Assuming this path must support X, should we …?”
- “This changes Y because …”

Avoid:

- Praise padding or performative politeness
- Blame language
- Vague statements such as “this is weird” or “clean this up”
- Calling an optional simplification broken
- Stating an assumption as fact
- Speculative future scenarios without a direct current risk
- Long essays when one concrete comment is sufficient

## Severity and Review Event

From OrestesK/omp, MIT.

More from OrestesK/omp

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Oct 2, 2026 (28 days ago)

From GitHub (fetched Oct 24, 2026).

Details

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

More from OrestesK

Cite this page

github-pr-comments skill. Agent Mart. https://agentmart.dev/skills/orestesk-github-pr-comments (last updated 2026-10-24)