Skip to content
Agentmart
Submit a tool

review

Fix-first review pipeline.

Skill by aitechnerd

IndexedMIT★ 1
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/aitechnerd-review

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

Fix-first review pipeline. Code reviewer auto-fixes mechanical issues and only escalates judgment calls. Run on any branch without building. Use /review [specific focus area].

Use cases: data

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/review

What's inside

Skill name
review
Description
Fix-first review pipeline. Code reviewer auto-fixes mechanical issues and only escalates judgment calls. Run on any branch without building. Use /review [specific focus area].
License
MIT
Instructions
96 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)

# Review — Fix-First Review Pipeline

## Overview

Runs the review pipeline with **fix-first** approach: the code reviewer
auto-fixes obvious mechanical issues (missing error handling, debug
statements, dead code, etc.) and only presents judgment calls for your
input. No SOW, no plan, no building — just review and fix.

## Step 1: Detect Changes

```bash
# Get the diff stats
git diff main...HEAD --stat
git diff main...HEAD --name-only
git log main..HEAD --oneline
```

If no changes compared to main:
> "No changes detected on this branch compared to main. Nothing to review."

## Step 2: Triage

Spawn **triage** (Haiku):
> "MODE: diff-triage. Summarize the git diff (main...HEAD) for this branch.
> Focus on: what changed, which files, what areas of the codebase.
> Note any security-relevant changes (auth, input handling, API, DB queries)."

If scanners are configured in `.claude/stack.md`:
```bash
bash ~/.claude/scripts/run-scanners.sh "/tmp/review-scans"
```

Spawn **triage** (Haiku):
> "MODE: scan-triage. Read reports in /tmp/review-scans/."

## Step 3: Code Review (Fix-First)

Spawn **code-reviewer** (Sonnet):
> "Review all changes on this branch (diff against main).
> Stack: read .claude/stack.md for language-specific checks.
> Triage summary: {paste diff-triage output}.
> $ARGUMENTS context: {user's focus area if provided}.
>
> **Fix-first mode:** Auto-fix all mechanical issues (missing error handling,
> debug statements, unused imports, dead code, obvious security fixes, etc.)
> with atomic commits. Report what you fixed and what needs human input.
>
> Save to /tmp/review-report.md."

## Step 4: DevSecOps — CONDITIONAL

**Skip if:** triage says SECURITY_RELEVANT = NO and scans clean.

Spawn **devsecops** (Sonnet) MODE 2:
> "Security review for branch changes. Triage: {scan-triage output}.
> Save to /tmp/security-review.md."

## Step 5: Present Results

> "**Review complete for branch `{current_branch}`**
>
> **Changes reviewed:** [N files, +X/-Y lines]
> **Auto-fixed:** [N issues] (mechanical fixes committed)
> **Needs your input:** [N issues] (judgment calls)
> **Security:** [CLEAN / N findings] (or skipped)
>
> **Auto-fixed items:**
> [list each: file:line — what was fixed — commit sha]
>
> **Needs input:**
> [for each ASK item: file:line — issue — options A/B/C — recommendation]
>
> [Show security findings if any]"

## Step 6: Apply User Decisions

If there are ASK items and the user responds:
- For items the user wants fixed → apply the recommended fix, commit
- For items the user skips → no action

If no ASK items remain (everything was auto-fixed):
> "All issues were mechanical and have been auto-fixed. You're good to ship."

---

## Notes

- This does NOT require a feature to be active. Works on any branch.
- Does NOT generate SOW, technical plan, or feature docs.
- Auto-fix commits use format: `fix(review): <what> at <file:line>`
- Good for: manual coding sessions, prototypes, other AI tool output,
  pre-PR sanity checks, code from new contributors.
- Stack profiles from `.claude/stack.md` are used if available.

From aitechnerd/ai-dev-team, MIT.

More from aitechnerd/ai-dev-team

Signals

Stars
1
Forks
0
Open issues
0
Last commit (this folder)
Mar 22, 2026 (7 months ago)

From GitHub (fetched Oct 24, 2026).

Details

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

More from aitechnerd

Cite this page

review skill. Agent Mart. https://agentmart.dev/skills/aitechnerd-review (last updated 2026-10-24)