Shweta-Mishra-ai/tokenmizer

Overview

Tokenmizer provides graph-based memory for AI agents, storing decisions, context, and session history that persist across conversations. It integrates with any LLM and is implemented in Python.

Best for

Best for
Developers experimenting with persistent memory for LLM agents in Python

Use cases

• Building conversational agents that retain long-term context
• Creating multi-session AI workflows with persistent decision history
• Enhancing LLM applications with structured memory beyond simple chat logs

How to use

Install

pip install "tokenmizer[anthropic,cache]"

Tested with

Claude Desktop, Claude Code, Cursor, Continue, VS Code, ChatGPT

Notes

Tokenmizer provides graph-based memory for AI agents, storing decisions, context, and session history that persist across conversations. It integrates with any LLM and is implemented in Python.

10 stars on GitHub. Last updated 2026-07-11. Licensed MIT.

Use cases

• Building conversational agents that retain long-term context
• Creating multi-session AI workflows with persistent decision history
• Enhancing LLM applications with structured memory beyond simple chat logs

Pros

• Works with any LLM, offering flexibility
• Graph memory structure enables complex context relationships
• Open source with a permissive license

Cons

• Very early stage with only 10 GitHub stars
• Limited community support and documentation
• No clear integration examples or benchmarks provided

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.