The Problem
Writing tools want to help you say what you mean. But the most interesting creative breakthroughs happen when you discover what you didn’t mean — or didn’t know you meant.
Most AI writing assistants correct, polish, and optimize. They make your writing more precise. But precision isn’t always what you need. Sometimes you need surprise.
The Approach
First Misread takes your text and deliberately gets it wrong — but in productive ways. It reads your words through lenses you wouldn’t choose: different emotional registers, unexpected metaphors, contradictory interpretations.
The misreadings aren’t random. They’re structured to surface:
- Buried assumptions you didn’t notice you were making
- Emotional undertones your rational mind filtered out
- Alternative framings that recast your argument entirely
- Creative tangents that lead somewhere unexpected
What I Built
A multi-persona system powered by Claude, where each “editorial persona” has a distinct misreading style:
- The Contrarian — reads everything as its opposite
- The Literalist — takes metaphors at face value
- The Poet — finds rhythm and imagery in dry prose
- The Therapist — reads between the lines for what you’re really saying
- The Historian — places your words in unexpected historical context
Available as a Claude Code skill, Obsidian plugin, and Chrome extension.
Tech Stack
- Core: Claude API with structured persona prompts
- Obsidian Plugin: TypeScript, Obsidian API
- Chrome Extension: TypeScript, Chrome Extensions API
- Orchestration: Claude Code skills and commands
How It Fits
First Misread is the creative counterpart to conventional AI writing tools. Where Grammarly makes you clearer, First Misread makes you more interesting — to yourself.
It’s become a core part of my writing workflow: draft → First Misread → discover what I actually wanted to say → redraft.