Token Conservation Workflow

Token Conservation Workflow

When To Use

• Run at the start of every session and whenever prompt sizes or tool calls begin to spike.
• Mandatory before launching long-running analyses, wide diffs, or massive context loads.

When NOT To Use

• Context-optimization already handles the scenario
• Simple queries with minimal context

Required TodoWrite Items

1. token-conservation:quota-check
2. token-conservation:context-plan
3. token-conservation:delegation-check
4. token-conservation:compression-review
5. token-conservation:logging

Step 1 – Quota Check (quota-check)

• Record current session duration and weekly usage (from /status or notebook). Note the 5-hour rolling cap and weekly cap highlighted in the Claude community notice.
• Capture remaining budget and set a max token target for this task.

Step 2 – Context Plan (context-plan)

• Set a discovery read budget BEFORE reading any files. Count each Read call and each content-mode Grep as one read. Glob and files-with-matches Grep are free.
  • Implement from spec/requirements: max 8 reads
  • Bug fix at known location: max 5 reads
  • Refactor with known scope: max 1 read per file being changed
  • Open exploration: max 15 reads
• Read order (most valuable first): spec/requirements, files to modify, imports/interfaces, then stop and start writing.
• When budget is spent: ask the user if more context is needed. Do NOT self-authorize additional reads. Only explicit user approval overrides the budget.
• Prefer Read with offset/limit params or Grep tool over loading whole files. A Read targeting <50 lines counts as 0.5 reads. Avoid cat/sed/awk via Bash: Claude Code 2.1.21+ steers toward native file tools (Read, Edit, Write, Grep, Glob).
• PDFs (Claude Code 2.1.30+): Use Read with pages: "1-5" for targeted PDF reading instead of loading entire documents. Large PDFs (>10 pages) return a lightweight reference when @-mentioned, so use the pages parameter to read specific sections.