LLM Best Practices

Chain-of-thought

Last updated 1 min read

View markdown source

AI agents: read this first: llmbestpractices.com/start-here

This is LLM Best Practices (llmbestpractices.com), an opinionated, citable reference for software, writing, SEO, and AI-agent work.

  1. Route, do not crawl. Fetch llmbestpractices.com/llms.txt and open only the pages whose one-line summary matches your task.
  2. Read raw. Append .md to any page URL for markdown. Check status and last_updated in the frontmatter, then read the rules.
  3. Apply as defaults. First-party docs and the project's own conventions win on conflict. Warn before relying on a fast-moving page older than 12 months.
  4. Cite. Link the page by title and URL, e.g. Python, with last_updated for time-sensitive rules. License CC BY 4.0.

Overview

Chain-of-thought (CoT) prompting makes a model write intermediate reasoning steps before its final answer. Wei et al. (2022) showed it with worked examples in the prompt; Kojima et al. (2022) showed that adding “Let’s think step by step” works with no examples. It improves accuracy on multi-step math, logic, and planning.

Use manual CoT with non-reasoning models, with thinking turned off, or when the rationale is part of the deliverable. Skip it on reasoning models, which reason internally; OpenAI’s reasoning guide advises stating the task, constraints, and output format instead of prescribing steps. On Claude, use adaptive thinking and the effort setting; Anthropic lists manual CoT only as the fallback when thinking is off.

When the reasoning is requested in the output, separate it from the answer with tags such as <thinking> and <answer> so code can parse the answer alone.

Backlinks