tokenbank
Token Bank — the local LLM gateway that sits between your AI agents and every provider. Know where tokens go · Spend less with smart routing to Ollama, Groq, GitHub Models · Earn by sharing idle quota on a community P2P network. One-click onboarding for Cursor, Claude Code, Codex CLI, Gemini CLI — no agent changes. Full trace, seamless model swap
一句话定位
Token Bank — the local LLM gateway that sits between your AI agents and every provider. Know where tokens go · Spend less with smart routing to Ollama, Groq, GitHub Models · Earn by sharing idle quota on a community P2P network…
安装方式
在 DSH 的 web profile 中执行下面的命令;装完刷新页面即可。
$ dsh plugin --profile web add 'git+https://github.com/wink-run/tokenbank.git#02b85992dad2a1df7a7962b173c0f6f956e3b294'
兼容的 DSH 版本
未声明
作者与热度
wink-run · ★ 110 · 更新于 2026-11-09
关于这个插件
Token Bank ## Your AI needs a manager. Personal AI hub · Your token manager. One local gateway for Claude Code, Codex, Cursor, WorkBuddy, Kimi Code and more — see every token, spend less, keep every AI asset in one place, orchestrate agents, and rent out what sits idle. Open source · Keep your workflow · One-click revert
Website · Free download · ▶ Watch on YouTube · 中文文档 · Architecture · Privacy Policy Overview
Token Bank is a next-generation personal AI resource hub that enables one-click onboarding and intelligent orchestration of mainstream AI tools like Claude, Cursor, Codex, and WorkBuddy through a local gateway architecture. Core Value Proposition Usage Transparency: Full-chain trace makes every token consumption accountable Cost Optimization: Smart routing automatically switches between local models, free quotas, paid subscriptions, and community-shared compute with lossless protocol adaptation Sharing Economy: P2P compute-sharing network creates a decentralized exchange for models and agents, monetizing idle resources into credits Technical Highlights
Zero-Intrusion Integration Declarative application handlers (CLI env injection + config hot-patching) enable seamless onboarding without modifying agent applications.
Multi-Protocol Adaptation Layer Transparent protocol conversion (Anthropic Messages, OpenAI Chat, Codex Responses) allows agents to use third-party models without awareness.
Unified Asset Layer Architecture Community agents run directly on users' existing agent applications (Codex, Claude, Cursor, etc.) without rebuilding harnesses, executing within user-accumulated MCP/Skill/Prompt assets for dual reuse of runtimes and tool ecosystems.
Scenario Routing Engine Routing policy learning from usage patterns Lossless context compression Vision enhancement layer for non-multimodal models (automatic image recognition injection)
MCP Built-in Relay & Resource Projection Gating Constructs personal knowledge and tool ecosystems with controlled resource deployment.
AI-Native Architecture Abandons traditional hard-coded rules; lets agents dynamically construct core capabilities (asset discovery, personalized recommendations, routing optimization) based on actual scenarios and continuously evolve—building an agent management platform with agents—achieving high flexibility and robustness.
Usage-Based Evolution The system automatically extracts work portraits from real call records and session patterns, driving personalized recommendations for MCP/Skill/Prompt/Agent and continuous optimization of routing strategies. Multi-device usage aggregation, agent orchestration, and more make Token Bank truly smarter with you. Why Token Bank
Six AI headaches. One hub. Token Bank is your personal AI hub — your AI's manager. Plug Claude Code, Codex, Cursor, WorkBuddy, Kimi Code and more into one local gateway without changing how you work: see clearly, spend less, stay simple, keep your AI assets in order and let them get smarter with you, orchestrate a team of agents from one sentence, and earn from idle by renting out what you don't use.
Five pillars: Architecture
Implementation notes: Core capabilities: one-click onboarding · seamless model swap · full trace
相似插件
dsh-our-free-model
Our Free Model — 在 dsh 里装上这个插件即可,无需登录、注册或填 API Key,就能使用包括 DeepSeek V4.1 Flash、Kimi K3 在内的前沿模型,桌面端另有 EAC 协付渠道…
next-ai-draw-io
Draw and edit draw.io diagrams by chatting with an AI. Web app, desktop app, and an MCP server for Claude Code, Cursor,…
MemOS
Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill reuse, with…