Sonnet builds, Haiku swarms, Opus reviews: a tiered Claude Code workflow
Arindam_1729 · x · 2026-10-10
The author argues most people misuse Claude Code by burning Opus-level intelligence on tasks Haiku can handle in parallel. His setup: Sonnet runs the main session writing code and running tests, Haiku spawns parallel subagents for file search and doc fetching, and Opus acts as architect invoked at only three moments — before locking the plan, after the same test fails twice, and before shipping to review the full diff. Paired with Anthropic's experimental advisor tool, which lets Claude consult a stronger model at key moments with full conversation context; server-side, available to subscription and API accounts, not yet on Bedrock/AWS.
More from coding & agent
- Using Grok Bot as an agent, not a chatbot: it learned a news workflow from one demonstration — Arindam_1729 · 2026-10-10
- Dev let an AI agent run a subreddit for 5 weeks: 30 posts, zero comments — sujingshen · 2026-10-10
- OpenAI explains why Codex defaults to 300k context instead of 1M — prd_008 · 2026-10-10
- Open-source REA hooks reverse engineering into Claude Code, Codex and Cursor via MCP — dotey · 2026-10-10
- Open-source self-hosted auth layer gives personal AI agents unified tool access — vaibhavbetter · 2026-10-10
- Anthropic's brief overage-continuation policy causes subagent start-stop loops — willcb · 2026-10-10