…
How this works. The tool runs on your machine, driven by your own Claude Code or Codex install, billed to your own subscription. This page sends the job; a tiny daemon you install does the spawning. Every run is read-only and sandboxed: the agent may call the trial tool and nothing else. There is no API key anywhere, so there is nothing of yours for us to leak or bill.
Claude Code and Codex are detected automatically. Gemini CLI works with the full version after purchase — it registers as a standard MCP server in your Gemini settings.
1 · get your install code
2 · connect your agent
No terminal needed. Add this connector link inside Claude (claude.ai: Settings → Integrations → Add custom connector — or the same setting in the desktop app):
Then just talk to it: paste some of your writing and a draft, and ask Claude to score the draft with the idiolect tools. Each score uses one free run.
Run it on your own machine instead (Claude Code or Codex installed, sandboxed):
or run it and type the code when asked:
waiting for the daemon to check in…
3 · run it
that was the trial
you own this
The full version: no sample cap, no watermark, the measurement itself
(describe), and a persistent voice profile on your machine.
Drop these next to your agent and it works everywhere — Claude Code,
Codex, Gemini CLI.