Fixing Verbose Claude: A Strict CLAUDE.md Prompt Contract
OutlawWizard · reddit · 2026-08-10
The author complains that Claude models (like Opus) tend to be significantly more verbose than GPT models, responding in lengthy paragraphs rather than getting straight to the point. To alter this default behavior, the author added a hard style contract to their CLAUDE.md file.
Key rules of this contract include:
- Open with the verdict: The very first sentence must state the conclusion (e.g., "Fixed.", "Root cause:"), never restating the question or context.
- Hard word ceiling: Target 100 words, with a hard limit of 250. Past 250, cut content rather than compressing prose.
- Banned elements: Prohibits Markdown headers, bold lead-ins, tables, emojis, pleasantries like "Great question", and step recaps in the final message.
- Structured output: Follows a verdict sentence with optionally one causal sentence, then flat bullets under bare labels (e.g., Changes:, Verification:). Each bullet is a single fact on one line.
- Inline evidence: Every claim must carry a clickable file link or the exact command run, avoiding restating file contents.
- Narrate while working: Progress and reasoning go into short notes between tool calls; the final message delivers only the result.
More from coding & agent
- DIRECT: retrieval-based next-question suggestions for conversational agents — burkov · 2026-10-16
- Claude Code and Codex now autocomplete the user's next action — tokenbender · 2026-10-16
- Hermes Agent ships official Reverse Engineer Anything plugin with one-command setup — Teknium · 2026-10-16
- MLExpert: 35B MoE Agent Trained to Autonomously Handle Machine Learning Engineering — Shangheng Du · 2026-10-16
- Cowork-Gym: Alibaba Synthesizes 64.9K Agent Work Environments, Boosting Qwen by 8.7 Points — Anni Zou · 2026-10-16
- MMSearch-OPD: Joint RL and On-Policy Distillation for Long-Horizon Multimodal Search Agents — PekingUniversity · 2026-10-16