simplify-code

Reviews recent code changes with four parallel subagents and applies safe cleanup fixes.

Updated Aug 21, 2026
One-click install
npx skills add https://github.com/ewtodd/son-of-anton --skill simplify-code-ewtodd
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: simplify-code
Source: https://github.com/ewtodd/son-of-anton/tree/main/skills/software-development/simplify-code
Command: npx skills add https://github.com/ewtodd/son-of-anton --skill simplify-code-ewtodd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Recent code changes often accumulate duplication, needless complexity, wasted work, and shallow band-aid fixes. Manually reviewing a diff for all of these quality dimensions at once is slow and easy to do superficially. ## Core Features & Use Cases - Parallel Four-Reviewer Fan-Out: Launches four focused reviewers (code reuse, code quality, efficiency, altitude) concurrently via delegate_task batch mode, each searching the full codebase for evidence. - Risk-Tiered Fix Application: Classifies findings as SAFE (auto-apply), CAREFUL (apply with test verification), or RISKY (flag for human review), then applies them in order with targeted test runs. - Flexible Scoping and Dry Run: Supports focus filters (reuse, quality, efficiency, altitude), dry-run report-only mode, and diff scoping by commit, staged changes, branch, or file. - Use Case: After finishing a feature branch, ask for a cleanup pass; the skill reviews the diff, deduplicates findings across reviewers, applies safe fixes like removing unused imports and flattening nested conditionals, and flags deeper architectural issues for your decision. ## Quick Start Ask the agent to simplify my recent changes and apply the safe fixes it finds.

Frequently Asked Questions about simplify-code

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I clean up my recent code changes automatically?▼

Ask the agent to simplify your changes. It captures the git diff, runs four parallel reviewers covering reuse, quality, efficiency, and altitude, then applies safe fixes like removing unused imports and commented-out code while running targeted tests to verify nothing breaks.

How does parallel code review with subagents work?▼

The skill uses delegate_task batch mode to launch four reviewers concurrently, each receiving the complete diff plus repo access. Each reviewer searches the codebase for evidence in its specialty area, then findings are merged, deduplicated, and applied by risk tier.

Can I run a code cleanup review without changing any files?▼

Yes, request a dry run by saying simplify but don't change anything. The four reviewers still run and present all findings across the SAFE, CAREFUL, and RISKY tiers, but nothing is applied until you approve.

What happens if subagent delegation is not available?▼

The skill falls back to an inline single-pass review where all four reviewer angles are worked through sequentially in the current context with the same standards. The final summary clearly states it was an inline review rather than the parallel fan-out.

When should I not use automated code simplification?▼

Avoid it for hunting correctness bugs, which belongs in a dedicated code review pass, and for large-scale refactoring of shared infrastructure. RISKY findings like public API renames or N+1 restructuring are flagged for human decision rather than auto-applied.