Claude Opus Optimization Skill

A Claude Code skill for maximizing the capabilities of Claude Opus, Anthropic's most powerful model, including advanced reasoning, complex code generation, multi-step analysis, and extended thinking patterns.

When to Use This Skill

Choose this skill when:

• Tackling complex multi-step reasoning or analysis tasks
• Generating or refactoring large, architecturally significant codebases
• Performing deep research synthesis across multiple domains
• Writing nuanced long-form content requiring sustained coherence
• Debugging intricate issues that require careful step-by-step analysis
• Working on tasks where accuracy matters more than speed

Consider alternatives when:

• Simple tasks where Haiku or Sonnet would suffice (save cost and latency)
• High-volume batch processing where speed matters more than depth
• Real-time chat applications requiring sub-second responses

Quick Start

Copy
# Configure Claude Code to use Opus for complex tasks
claude config set model claude-opus-4-20250514

# Run a complex analysis task
claude "analyze this codebase architecture and suggest improvements"

# Use extended thinking for difficult problems
claude "solve this algorithmic problem step by step with extended thinking"

Copy
import Anthropic from '@anthropic-ai/sdk';

const anthropic = new Anthropic();

// Opus with extended thinking for complex reasoning
const response = await anthropic.messages.create({
  model: 'claude-opus-4-20250514',
  max_tokens: 16000,
  thinking: {
    type: 'enabled',
    budget_tokens: 10000
  },
  messages: [{
    role: 'user',
    content: 'Analyze the time complexity of this algorithm and optimize it...'
  }]
});

Core Concepts

Extended Thinking

Feature	Description	Use Case
Thinking Budget	Allocate tokens for internal reasoning	Complex math, logic puzzles, multi-step analysis
Chain of Thought	Step-by-step visible reasoning	Debugging, architectural decisions
Self-Verification	Model checks its own work	Critical calculations, code correctness
Backtracking	Reconsider and revise approach	Ambiguous problems, creative solutions

Prompt Patterns for Opus

Copy
# Pattern: Expert Analysis
You are a senior software architect with 20 years of experience.
Analyze the following codebase considering:
1. Architectural patterns and anti-patterns
2. Scalability bottlenecks
3. Security vulnerabilities
4. Performance optimization opportunities
5. Maintainability and code quality

Provide your analysis in a structured report with severity ratings.

# Pattern: Multi-Perspective Review
Consider this problem from three perspectives:
- **Engineering**: Technical feasibility and implementation cost
- **Product**: User impact and business value
- **Operations**: Deployment risk and maintenance burden

Synthesize these perspectives into a recommendation.

# Pattern: Systematic Debugging
Approach this bug systematically:
1. Reproduce: What exact steps trigger the issue?
2. Isolate: What is the minimal failing case?
3. Hypothesize: What are the three most likely root causes?
4. Verify: How would you test each hypothesis?
5. Fix: What is the safest, minimal fix?

Configuration

Parameter	Type	Default	Description
model	string	"claude-opus-4-20250514"	Fixed to Opus for this skill
max_tokens	number	16000	Maximum output tokens (Opus supports up to 32K)
thinking.type	string	"enabled"	Enable extended thinking for complex tasks
thinking.budget_tokens	number	10000	Token budget for internal reasoning
temperature	number	1.0	Output randomness (lower for analytical tasks)
system	string	""	System prompt for task-specific behavior
stop_sequences	array	[]	Custom stop sequences

Best Practices

1. Reserve Opus for tasks that genuinely need it — Opus excels at complex reasoning, large-scale code generation, and nuanced analysis; for simple tasks, Sonnet or Haiku delivers faster results at lower cost.

2. Use extended thinking for multi-step problems — enable the thinking budget when tasks require working through multiple steps, verifying intermediate results, or exploring alternative approaches before committing to an answer.

3. Provide rich context and clear success criteria — Opus benefits from detailed background information and explicit criteria for what constitutes a good answer; the more context you provide, the more its advanced reasoning can leverage.

4. Structure prompts with numbered steps or roles — break complex requests into explicit steps or assign a specific expert role; this helps Opus organize its reasoning and produce more thorough outputs.

5. Review thinking traces for insight into reasoning — when extended thinking is enabled, examine the thinking output to understand how the model approached the problem; this helps you refine prompts and catch reasoning errors.

Common Issues

Responses are slow for simple tasks — Opus prioritizes depth over speed. If your task is straightforward, switch to Sonnet or Haiku. Use Opus selectively for tasks where its advanced reasoning provides measurable value.

Extended thinking uses too many tokens — The thinking budget adds to your total token usage. Set budget_tokens to a reasonable limit (5000 for moderate tasks, 10000+ for complex ones) and monitor usage. Not every request needs the maximum thinking budget.

Output is overly detailed or verbose — Opus tends to be thorough, which can mean longer responses than needed. Add explicit length constraints in your prompt like "respond in under 500 words" or "provide a concise summary" to control output length.