Keep the signal. Cut the noise. Local CLI + MCP — fold verbose tool output for Claude Code, Codex, Cursor & Desktop.
Smaller tool outputs, zero LSP daemons, with omitted context available on demand. A local CLI, MCP server, and macOS desktop HUD for Claude Code, Codex / ChatGPT Desktop, Antigravity, and Cursor.
Try in 1 Min · macOS Desktop Pet · Zero-LSP Code Intel · Architecture · Integrations
Requires Python 3.11+ and uv. Install this repository: the PyPI name tokencut belongs to another project.
The demo works outside a project and makes no model calls. It measures an authored log, checks the complete diagnostic tail, verifies recovery of the original, and confirms that unfamiliar output stays unchanged. Checks print PASS or FAIL; a failed check returns a nonzero exit code. Use tokencut demo --json for machine-readable results. Demo data uses a disposable cache.
Then try your own command: tokencut run -- , or connect Claude Code, Claude Desktop, Codex, or Antigravity. Smaller tool output is the measured benefit; subscription limits and task quality need separate evaluation.
Illustrative terminal demo. Run the fixture benchmark below for reproducible measurements.
macOS desktop pet companion (local preview)
TokenCut Desktop Pet — A draggable macOS companion monitoring real-time tool output reduction and 5-hour model quota windows.
A small draggable SwiftUI pet lives above other windows, remembers its position, and keeps working across Spaces. Click it to see quota windows, reset times and estimated tool-output reduction, before/after counts, recovery costs, seven days of history and breakdowns by client, Git project and tool. The UI is in English, follows the system appearance,
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/00200200--tokencut.html)
Keep the signal. Cut the noise. Local CLI + MCP — fold verbose tool output for Claude Code, Codex, Cursor & Desktop.
kdev1966/Handoff-CLIShared, persistent memory for your AI coding assistants (Claude Code, Copilot, Cursor, Gemini CLI, Codex…): a local MCP server and CLI that hand work over between tools automatically.
thaildhe172591/pythiaPL/SQL development for AI agents on Oracle Database - expert queries, impact analysis, and a snapshot-verified write path with honest rollback
shinegang/baronBaron Munchausen — local memory for coding agents that outlives the chat, with a verdict on every answer. Public alpha.
dontpayfull/AInotateAnnotated screenshots for AI agents: Skitch-style arrows, numbered steps, boxes, labels, solid redaction
Farukes/TokenJarZero-cost token optimization engine for AI coding assistants (Claude Code, Cursor, Antigravity, Windsurf). Saves 70-95% tokens via AST skeletons, differential cache, and line slicing.
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.