6 tips to cut Claude Code workflow costs: effort levels, token budgets, loop caps
majidmanzarpour · x · 2026-10-01
A developer shares six practical tips for controlling token costs in Claude Code workflows (ultracode):
- Set effort first: /effort high or medium for routine work; request xhigh/max only for named hard parts.
- Use plain language instead of the keyword, with explicit limits (e.g., max 6 agents, effort high, one fix round plus one review).
- Set a token budget: '+2M tokens' becomes a hard ceiling the workflow can read and size itself to.
- Cap the loops: 'max 2 rounds' or 'stop after first approval' — open-ended loops are the biggest cost multipliers.
- Name what can be cheaper: mechanical steps (metrics, renders, docs) at low/medium; judgment steps (reviews, hard debugging) at high/xhigh.
- Ask for a checkpoint: pause after phase 1 and show results before continuing to prevent runaway fan-out.
Reserve the 'ultracode' keyword for tasks where exhaustiveness is worth the cost; plain workflows with limits give most of the benefit at a fraction of the price.
Related event: Six Tips to Cut Token Costs in Claude Code Workflows(2 posts)→
More from coding & agent
- HF researcher admits a well-maintained monorepo is the superior way to run an AI lab — soldni · 2026-10-01
- uv replaces five Python tools at 10-100x pip speed, written in Rust — blaizedsouza · 2026-10-01
- KV-streams trains SWE agents 2x faster by preserving KV cache across compaction — burny_tech · 2026-10-01
- LangChain uses 'human touches per PR' as a key metric to evaluate coding agents — LangChain · 2026-10-01
- LangChain dev shows Slack-triggered coding agent that auto-addresses PR comments — LangChain · 2026-10-01
- CorpusMap organizes growing corpora by recurring entities so agents follow links and save tokens — coallaoh · 2026-10-01