Claude Code Effort Loop

A systematic multi-stage workflow for building software with variable AI effort levels.

PlayableOpen source

Description

A verified workflow for using Claude Code that sequences low-effort implementation passes for initial feature development with high-effort verification steps for edge-case testing. The method leverages internal benchmarks like Terminal-Bench 3.0 to demonstrate how varying AI compute effort improves success in security, hardware, and software engineering domains.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: Provides a specific, benchmarked strategy for optimizing AI compute effort to solve complex, edge-case-heavy…
Built with: Opus 5.5 Fable 5.1
Links: claude.dev
View tweet
View GitHub repo
Tags: #workflow #coding #productivity #benchmarking

Discovered: X
First posted: X · Sep 25, 2026
Listed: Added 16d ago
Build: Strong
Models: Strong

Prompt

Set up the workflow in this image in my Claude Code:

1. Create two subagents in ~/.claude/agents:
- builder: writes the code. model: opus, effort: low. It implements the spec I've approved and stops when it's done.
- verifier: checks the work. model: opus, effort: high. It runs the feature first to see what it actually does, then adds tests, focusing on edge cases. It reports problems to me and doesn't change the code itself.

2. Add a rule to ~/.claude/CLAUDE.md: when I ask for a new feature, first ask me what the spec is missing; after I confirm, hand it to builder; after I've reviewed the result, hand it to verifier.

3. Check whether the environment variable CLAUDE_CODE_EFFORT_LEVEL is set. If it is, the effort settings on both subagents won't take effect. Just tell me; don't change it.

Show me the contents of the files you'll create or change, then explain the whole workflow in words a 10-year-old could follow. Don't write anything until I confirm.

Source

Explore

Agents / Skills