Claudeisburningthroughyourtokens.Watchithappen.
Wick sits quietly below your chat input. Real-time usage, model-aware estimates, and a one-click escape hatch to ChatGPT, Gemini, or Grok. Also runs as a local MCP server for Claude Desktop and Claude Code.
"You find out you've hit the limit when Claude stops responding. Never before."
Sees what Claude hides
A live meter below your input shows exact session and weekly usage - more precise than Claude's own usage page.
Knows your model
Opus burns 5× faster than Haiku. Wick weighs every estimate by the model you're actually on.
One click to escape
Hit a limit? Send your entire conversation - including files Claude built - to ChatGPT, Gemini, or Grok instantly.
Also lives in your terminal.
The extension meters claude.ai. The MCP server meters everything else - Claude Desktop, Claude Code, and CLI. One product. Every Claude surface.
Automatic. Wick reads the usage logs Claude Code already writes - real token counts, zero setup, no custom instructions.
One-time custom instruction lets Claude report each turn via the wick_track tool.
Zero setup for Claude Code - it reads the logs automatically. One custom instruction for Desktop.
A live dashboard. At localhost:6789.
Every turn tracked. Cost in rupees. Burns down in real time.
A single line.
Always there.
Under your chat input, not in a sidebar. Not in a separate tab. Right where you type.
"We burned through the 5-hour limit every afternoon without knowing why. Wick showed us in 10 minutes."
We never see your prompts.
Wick runs entirely in your browser. Usage counting happens locally. Your conversations never leave your machine unless you press a handoff button - and even then, they go directly to the destination, not through our servers.
Read the full privacy policy →Straight answers.
Stop flying blind.
Chrome extension in ninety seconds. MCP server in thirty.