code-reviewer

Code Reviewer and Documenter Skill

Purpose

Continuously monitor a codebase for changes and produce standardized per-commit documentation including architecture changes, features, regressions, and breaking changes.

Trigger

This skill should be invoked:

1. When user requests codebase analysis
2. Periodically to check for new commits
3. After any development session

Quick Check Command

# Check for new commits since last analyzed
git log --oneline HEAD ^LAST_ANALYZED_SHA

Output Location

All documentation goes into bot-analysis/ folder in the repository root.

Document Structure

bot-analysis/
+-- INDEX.md              # Main navigation (chronological commit table)
+-- README.md             # Quick start and status
+-- commits/
    +-- 01_XXXXXXX.md     # First commit analysis
    +-- 02_XXXXXXX.md     # Second commit analysis
    +-- ...
    +-- NN_XXXXXXX.md     # Latest commit (HEAD)

Per-Commit File Template

# Commit XXXXXXX - [Short Description]

| Field | Value |
|-------|-------|
| SHA | `XXXXXXX` |
| Timestamp | YYYY-MM-DD HH:MM:SS +ZZZZ |
| Message | [commit message] |
| Sequence | #N of M |
| Delta | +X minutes from previous |
| AI Model | [model name or Unknown] |

**Navigation:** [Index](../INDEX.md) | Prev: [XXXXXX](NN_XXXXXX.md) | Next: [XXXXXX](NN_XXXXXX.md)

---

## Status

| Aspect | State |
|--------|-------|
| Video | [Working/Broken/etc] |
| Audio | [Working/Broken/etc] |
| GUI | [Working/Broken/etc] |
| Tests | [X passing] |

---

## Architecture Changes

[Diff tree or "No changes"]

---

## Features Introduced

[New features or "None"]

---

## Breaking Changes

[Breaking changes or "None"]

---

## Regressions

[New regressions introduced or fixed, with ID like R-XXX]

---

## Notes

[Additional observations]

Continuous Monitoring Workflow

Step 1: Check for New Commits

git log --format="%h|%ai|%s" HEAD -1

Compare with last entry in commits/ folder.

Step 2: If New Commits Found

1. Get commit details:

   git show COMMIT_SHA --stat
   git diff PREV_SHA..COMMIT_SHA --name-status
   

2. Create new commit file: commits/NN_XXXXXXX.md

3. Update INDEX.md with new row

4. Update navigation links in previous commit file

Step 3: Analyze for Breaking Changes

Check for:

• Signature changes in exported functions/classes
• Removed or renamed files
• Changed input/output types
• Modified API contracts

Step 4: Update Regression Tracking

• Assign ID (R-XXX) to new regressions
• Link to commit that introduced them
• Update status when fixed

Key Principles

1. Objectivity: Reference SHA commits, not subjective descriptions
2. Time-Driven: Order by timestamp, from initial to latest
3. Navigatable: Use Prev/Next/Index links
4. Delta Focus: Emphasize what changed, not cumulative state
5. Breaking Changes: Always highlight API/interface changes
6. Non-Coding: Analyze only, do not modify source code

Regression ID Format

R-XXX: [Description]
Introduced: SHA (timestamp)
Fixed: SHA or UNRESOLVED
Impact: [affected components]

Status Values