ECC/rules/common/performance.md

Performance Optimization

Model Selection Strategy

Haiku (90% of Sonnet capability, 3x cost savings):

• Lightweight agents with frequent invocation
• Pair programming and code generation
• Worker agents in multi-agent systems

Sonnet (Best coding model):

• Main development work
• Orchestrating multi-agent workflows
• Complex coding tasks

Opus (Deepest reasoning):

• Complex architectural decisions
• Maximum reasoning requirements
• Research and analysis tasks

Context Window Management

Avoid last 20% of context window for:

• Large-scale refactoring
• Feature implementation spanning multiple files
• Debugging complex interactions

Lower context sensitivity tasks:

• Single-file edits
• Independent utility creation
• Documentation updates
• Simple bug fixes

Extended Thinking + Plan Mode

In Claude Code, adaptive reasoning lets the model choose how much to think for each task. For Fable models, Sonnet 5 and later, and Opus 4.7 and later, use effort to control reasoning rather than a fixed token budget. Available levels depend on the model.

Control thinking via:

• Effort: Use /effort in a session, --effort <level> at launch, or CLAUDE_CODE_EFFORT_LEVEL. Save a per-model level in modelSettings in ~/.claude/settings.json; the older user-level effortLevel key does not apply to Opus 5.5 or newer models.
• Toggle: Option+T (macOS) / Alt+T (Windows/Linux), or alwaysThinkingEnabled in settings. These cannot disable thinking on Opus 5.5, Sonnet 5.5, or Fable models.
• Fixed budget: Claude Code ignores positive MAX_THINKING_TOKENS values on adaptive reasoning models. On Opus 4.6 or Sonnet 4.6, set CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING=1 to use a fixed budget with MAX_THINKING_TOKENS.
• Verbose mode: Ctrl+O to see thinking output

See Claude Code model configuration and environment variables for supported controls.

For complex tasks requiring deep reasoning:

1. Choose a higher supported effort level, such as /effort high
2. Enable Plan Mode for structured approach
3. Use multiple critique rounds for thorough analysis
4. Use split role sub-agents for diverse perspectives

Build Troubleshooting

If build fails:

1. Use build-error-resolver agent
2. Analyze error messages
3. Fix incrementally
4. Verify after each fix