Tokenmizer
- 23 repo stars
- Updated July 27, 2026
- Shweta-Mishra-ai/tokenmizer
TokenMizer is a local LLM proxy and Claude Code plugin that applies graph memory, checkpointing, and compression so developers can keep context across long sessions.
About
TokenMizer is a Claude Code marketplace plugin backed by a local LLM proxy that never lets you lose AI context again, per its manifest. It adds graph-backed memory, session checkpointing, file intelligence, compression, and caching so hours of decisions survive context limits. Developers reach for it when repeated re-explaining or huge files are burning tokens and breaking continuity across sessions.
- Graph-backed memory and session checkpointing
- Local proxy for Claude, GPT, Gemini, and other LLMs
- File intelligence and compression pipeline
- MCP-compatible Claude Code plugin
Tokenmizer by the numbers
- Data as of Jul 28, 2026 (Skillselion catalog sync)
/plugin marketplace add Shweta-Mishra-ai/tokenmizerAdd your badge
Show developers this marketplace is listed on Skillselion. Paste this into your README.
| repo stars | ★ 23 |
|---|---|
| Last updated | July 27, 2026 |
| Repository | Shweta-Mishra-ai/tokenmizer ↗ |
How do you stop losing decisions and rationales when context windows fill or sessions restart?
Proxy LLM calls through graph memory, compression, caching, and checkpoints so long coding sessions keep context.
Who is it for?
Developers running Claude Code or other LLM clients who need durable session memory and token-efficient proxies.
Skip if: Projects that send only short, stateless prompts with no multi-session continuity needs.
What you get
Checkpointed sessions, compressed inputs, cached repeats, and a live knowledge graph feeding the next request.
Plugins in this marketplace
1 plugin - install individually after you add the marketplace.
Recommended Marketplaces
FAQ
What problem does TokenMizer address?
It targets context loss when limits hit, including re-explaining work and large files consuming the budget.
How does traffic flow?
Your app talks to TokenMizer on port 8000, which runs the pipeline then forwards to Claude, GPT, Gemini, or other providers.