Advice for Working Around Peak Hours¶

Peak hours mean tighter session limits, not less total capacity. With a bit of planning, you can get the same amount of work done without hitting walls.

Plan your day around the schedule¶

Shift heavy work to off-peak¶

Token-intensive tasks burn through your session limits fast during peak. Move them to off-peak windows:

• Large refactors spanning many files
• Code generation for new features or boilerplate
• Background agents running autonomously
• Batch operations like bulk file processing or migrations

"If you run token-intensive background jobs, shifting them to off-peak hours will stretch your session limits further."

Reduce token usage during peak¶

When you do work during peak hours, make every token count:

• Keep prompts focused. Ask one thing at a time instead of multi-part requests.
• Use /compact to compress conversation context when it grows large.
• Disable unused plugins. Each active MCP server or skill adds to context size.
• Don't dump entire files. Point Claude to specific lines or functions instead of pasting whole files.
• Start fresh sessions. A clean conversation uses fewer tokens than continuing a long one.

Use peak hours for low-token tasks¶

Some work barely touches your limits. Save these for peak hours:

• Code review — reading diffs, spotting issues
• Planning — designing architecture, writing specs
• Quick questions — short Q&A about APIs or syntax
• Documentation — writing or editing docs
• Manual work — tasks you do yourself with Claude on standby

Spread work across tools¶

Diversifying across AI tools means no single one bottlenecks you:

• Google Gemini — strong for code and general tasks
• OpenAI ChatGPT / Codex — alternative coding assistant
• Local models (Ollama, LM Studio) — unlimited, no rate limits
• GitHub Copilot — inline completions don't count against Claude limits

Use Claude for what it does best and fall back to alternatives when you're near your limit.

Weekly rhythm¶

Your weekly token budget doesn't change — only how it's distributed. Think of it like electricity pricing: same total supply, cheaper at off-peak times.

• Monday-Friday mornings (PT): conserve, do lightweight work
• Afternoons and evenings: go heavy, no restrictions
• Weekends: full capacity, great for deep work sessions

The weekly schedule page shows exactly where you are right now.