Turning Claude Code into a Writing Harness with Claude Mods and an Output Style
Writing articles in Claude Code has advantages that chat on claude.ai doesn't. You can build your own harness of skills and rules, and you have a lot of freedom over what context Claude gets. I write my articles in this repository (zenn-content), where I keep skills for writing and agents for review. But even when I opened the writing repository, what Claude saw was the global harness I had grown…
Claude Code offers advantages for article writing compared to Claude on claude.ai. This allows for building custom harnesses with rules and instructions tailored to specific tasks. The article discusses using Claude Mods and output styles to create a writing harness. Mods let you control what Claude sees, while the output style determines Claude's conversational style.
The Mod integrates with Claude Code's skill listing and instruction files. By measuring the impact of turning off certain skills and agents, the article demonstrates how this affects Claude's context window and character budget. A profile-based system called harness-scope is introduced to easily manage these customizations. Applying the zenn-writing profile, which removes specific skills, agents, and instruction files, significantly reduced the size of the skill listing and agent list.
This approach allows for a highly specialized writing harness tailored to the needs of producing articles, while still leveraging Claude Code's powerful capabilities.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
Also reported by 1 other outlet
- Anthropic opens Claude Code to internal Mods thearabianpost.com
