Code Optimizer

Code optimizer (three subagents)

When to use

• Performance problems or slow code
• Suspected memory leaks or resource misuse
• Silent bugs or edge cases
• Deep code review before release

Goal

Identify and fix bugs, inefficiencies, and resource issues using a three-stage pipeline with required user confirmation before applying changes.

Pipeline overview

Parent Agent
  -> Subagent 1: Analyzer (write /tmp/code-analysis.md)
  -> Subagent 2: Planner (write /tmp/optimization-plan.md)
  -> Pause for user approval
  -> Subagent 3: Executor (apply fixes, write OPTIMIZATION_REPORT.md)

Phase 1: Parent reconnaissance

1. Determine scope (files, directory, or full codebase)
2. Collect context (CLAUDE.md, package.json, requirements.txt, etc.)
3. Run baseline tests if present
4. Pass scope and context to Subagent 1

Phase 2: Subagent 1 (Analyzer)

Use the analysis checklist and output format in [references/analysis-checklist.md](references/analysis-checklist.md). Write /tmp/code-analysis.md.

Phase 3: Subagent 2 (Planner)

Use the plan template in [references/plan-format.md](references/plan-format.md). Write /tmp/optimization-plan.md.

Phase 4: Pause and confirm

Present the plan summary and ask the user whether to apply all fixes, only critical/high, or a subset. Do not continue without explicit confirmation.

Phase 5: Subagent 3 (Executor)

Apply only approved fixes, run tests for high-risk changes, and write OPTIMIZATION_REPORT.md.

Notes

• Subagents 1 and 2 never modify files.
• The pause before execution is mandatory.
• This skill complements refactoring (structure/readability) with correctness and performance focus.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Mingo07-dev
• Source: Mingo07-dev/claude_skills
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.