diff --git a/mintlify-docs/ai-tools/faq.mdx b/mintlify-docs/ai-tools/faq.mdx index 594de22..f902ae9 100644 --- a/mintlify-docs/ai-tools/faq.mdx +++ b/mintlify-docs/ai-tools/faq.mdx @@ -41,7 +41,7 @@ description: "Common questions about the KaleidoSwap AI tools, covering custody, It varies by surface: - **KaleidoAgent** reasons with a hosted model, Claude or OpenAI, selected with `AGENT_PROVIDER`. - - **KaleidoMind** runs the model **on the device** through the QVAC SDK. The recommended model is Qwen3.5 2B, on desktop and phone alike. Its tiered funnel means most requests never reach the model at all. Pairing a phone with the desktop app for delegated inference is paused in desktop app v0.5.1. + - **KaleidoMind** runs the model **on the device** through the QVAC SDK. The recommended model is Qwen3.5 2B, on desktop and phone alike. Its tiered funnel means most requests never reach the model at all. Inference always runs on the device you are using; offloading to another device is not available yet. - **MCP servers** are model-agnostic. Whatever model your MCP host runs is the one calling the tools. diff --git a/mintlify-docs/ai-tools/installation.mdx b/mintlify-docs/ai-tools/installation.mdx index b36e16b..a24d790 100644 --- a/mintlify-docs/ai-tools/installation.mdx +++ b/mintlify-docs/ai-tools/installation.mdx @@ -68,7 +68,7 @@ KaleidoMind is an engine embedded in a host application rather than a standalone ### Through the Desktop App -The fastest path. The [Desktop App](/desktop-app/getting-started/installation) bundles the engine as in-app chat, manages the on-device model lifecycle, and can act as the paired inference peer a phone delegates to. Switch it on from **Settings > Capabilities** by selecting **Node + Mind** or **Only Mind**. +The fastest path. The [Desktop App](/desktop-app/getting-started/installation) bundles the engine as in-app chat, and manages the on-device model lifecycle. Switch it on from **Settings > Capabilities** by selecting **Node + Mind** or **Only Mind**. KaleidoMind depends on a local AI runtime and may not be available on Windows yet. Every trading and node feature works without it, so run the app in **Only Node** mode if Mind is unavailable on your platform. @@ -105,4 +105,4 @@ To exercise the engine against a model without a phone, use the playground from pnpm play "pay bob 3 eur" ``` -For how the tiered funnel routes a request, the tool contract shared across transports, the skills the engine loads, and where it runs across desktop, phone, and paired devices, check the [KaleidoMind page](/ai-tools/kaleido-mind). +For how the tiered funnel routes a request, the tool contract shared across transports, the skills the engine loads, and where it runs on desktop and phone, check the [KaleidoMind page](/ai-tools/kaleido-mind). diff --git a/mintlify-docs/ai-tools/kaleido-mind.mdx b/mintlify-docs/ai-tools/kaleido-mind.mdx index 17684f7..b97841e 100644 --- a/mintlify-docs/ai-tools/kaleido-mind.mdx +++ b/mintlify-docs/ai-tools/kaleido-mind.mdx @@ -20,7 +20,7 @@ Most requests never reach the model at all. |------|---------|------| | `T0` fast path | "balance", "address", "btc price" | Zero inferences, instant | | `T2` recipe | "pay bob 3 EUR", "buy 0.001 BTC" | About one inference (the model may assist slot extraction), then a deterministic chain, confirm-gated | -| `T1` agentic loop | Everything else | Skill-scoped LLM, can P2P-delegate hard or novel chains to a paired desktop's bigger model | +| `T1` agentic loop | Everything else | Skill-scoped LLM on the local model | - **T0, fast path.** Deterministic pattern match, no inference. Balance checks, addresses, spot prices. - **T2, recipe engine.** A skill carries the ordered plan (resolve, price, convert, confirm, send); the model only fills the slots. That makes multi-step flows reliable even on a ~0.6B parameter model, instead of asking the model to plan the whole chain itself. @@ -32,7 +32,7 @@ The model sees identical tool names and schemas everywhere, only *how a tool exe | Surface | Tool execution | Confirm before spend | |---------|----------------|----------------------| -| Mobile (Rate) | In-process WDK adapters (fully on-device, private); P2P-delegate to a paired desktop optional | Confirmation sheet | +| Mobile (Rate) | In-process WDK adapters (fully on-device, private) | Confirmation sheet | | Desktop | `kaleido-mcp` consumed as a stdio tool source (namespaced `spark_*`/`rln_*`/`kaleidoswap_*`) | Confirmation dialog | | Eval / CLI | Canned stub handlers for reproducible benchmarks | Auto-approve (asserts the gate fired) | @@ -61,11 +61,11 @@ Skills are Agent-Skills-spec playbooks (`SKILL.md` plus progressive disclosure) ## QVAC: On-Device Inference -LLM, embedding, speech-to-text, and text-to-speech inference all run through the [QVAC SDK](https://www.npmjs.com/package/@qvac/sdk), Tether's local AI runtime, locally on-device by default, or delegated to an explicitly paired, user-controlled desktop for heavier work. Published as the `@kaleidorg/mind/qvac` subpath so the SDK stays a peer dependency rather than a hard requirement of core. Memory and RAG (long-term recall, wallet-history retrieval, merchant discovery) also route through QVAC's embeddings, with near-duplicate consolidation to keep memory from bloating. +LLM, embedding, speech-to-text, and text-to-speech inference all run through the [QVAC SDK](https://www.npmjs.com/package/@qvac/sdk), Tether's local AI runtime, locally on the device. Offloading inference to another device is planned but not available yet. Published as the `@kaleidorg/mind/qvac` subpath so the SDK stays a peer dependency rather than a hard requirement of core. Memory and RAG (long-term recall, wallet-history retrieval, merchant discovery) also route through QVAC's embeddings, with near-duplicate consolidation to keep memory from bloating. | Package | Version | |---------|---------| -| `@qvac/sdk` | The core engine works with 0.13.1 and later and is tested with 0.21. P2P delegation needs 0.13.1–0.18: QVAC removed it in 0.19 | +| `@qvac/sdk` | The core engine works with 0.13.1 and later and is tested with 0.21 | | `@kaleidorg/mind` | Takes `@qvac/sdk` as an optional peer dependency, so the engine also runs with no model at all against the mock provider | ### Recommended models @@ -102,7 +102,7 @@ For a step-by-step version on signet, follow [Build a Local RGB Agent](/ai-tools | Host | Role | |------|------| | [Rate](https://github.com/kaleidoswap/Rate) | React Native mobile wallet, with local LLM, speech-to-text, neural text-to-speech, and the hands-free voice loop | -| [Desktop App](/desktop-app/getting-started/introduction) | Runs the engine as in-app chat through the Tauri sidecar (`apps/provider`), which consumes `kaleido-mcp` as a stdio tool source, and can act as the paired inference peer for a phone (phone pairing is paused in desktop app v0.5.1) | +| [Desktop App](/desktop-app/getting-started/introduction) | Runs the engine as in-app chat through the Tauri sidecar (`apps/provider`), which consumes `kaleido-mcp` as a stdio tool source | The fastest way to try it is the Desktop App. [Download the latest release](https://kaleidoswap.com/downloads) and follow the [installation guide](/desktop-app/getting-started/installation). diff --git a/mintlify-docs/cn/ai-tools/faq.mdx b/mintlify-docs/cn/ai-tools/faq.mdx index 403e22e..1e9b588 100644 --- a/mintlify-docs/cn/ai-tools/faq.mdx +++ b/mintlify-docs/cn/ai-tools/faq.mdx @@ -41,7 +41,7 @@ description: "关于 KaleidoSwap AI 工具的常见问题：托管方式、种 因接入方式而异： - **KaleidoAgent** 用托管模型推理，Claude 或 OpenAI，通过 `AGENT_PROVIDER` 选择。 - - **KaleidoMind** 通过 QVAC SDK 让模型**在设备上**运行。推荐使用 Qwen3.5 2B，桌面端和手机都适用。它的分层漏斗设计意味着大多数请求根本不会到达模型。桌面应用 v0.5.1 中，手机与桌面配对进行委托推理的功能已暂停。 + - **KaleidoMind** 通过 QVAC SDK 让模型**在设备上**运行。推荐使用 Qwen3.5 2B，桌面端和手机都适用。它的分层漏斗设计意味着大多数请求根本不会到达模型。推理始终在你正在使用的设备上运行，暂不支持把推理转移到其他设备。 - **MCP 服务器** 与模型无关。调用工具的就是你的 MCP 宿主所运行的那个模型。 diff --git a/mintlify-docs/cn/ai-tools/installation.mdx b/mintlify-docs/cn/ai-tools/installation.mdx index d55fe0f..7e7dab8 100644 --- a/mintlify-docs/cn/ai-tools/installation.mdx +++ b/mintlify-docs/cn/ai-tools/installation.mdx @@ -68,7 +68,7 @@ KaleidoMind 是嵌入宿主应用中的引擎，而不是一个独立安装的 ### 通过桌面应用 -这是最快的路径。[桌面应用](/cn/desktop-app/getting-started/installation)将该引擎打包为应用内聊天，负责管理本地模型的生命周期，并且可以作为手机端委托推理的配对节点。在 **设置 > 能力** (Settings > Capabilities) 中选择 **Node + Mind** 或 **Only Mind** 即可开启。 +这是最快的路径。[桌面应用](/cn/desktop-app/getting-started/installation)将该引擎打包为应用内聊天，并负责管理本地模型的生命周期。在 **设置 > 能力** (Settings > Capabilities) 中选择 **Node + Mind** 或 **Only Mind** 即可开启。 KaleidoMind 依赖本地 AI 运行时，目前在 Windows 上可能尚不可用。所有交易与节点功能都不依赖它，因此如果你的平台上 Mind 暂不可用，可以用 **Only Node** 模式运行应用。 @@ -105,4 +105,4 @@ npx tsx src/index.ts skills # 列出已安装的 Skills pnpm play "pay bob 3 eur" ``` -分层漏斗如何路由一次请求、跨传输方式共享的工具契约、引擎加载的 skills，以及它在桌面、手机和配对设备上的运行方式，请查看 [KaleidoMind 页面](/cn/ai-tools/kaleido-mind)。 +分层漏斗如何路由一次请求、跨传输方式共享的工具契约、引擎加载的 skills，以及它在桌面和手机上的运行方式，请查看 [KaleidoMind 页面](/cn/ai-tools/kaleido-mind)。 diff --git a/mintlify-docs/cn/ai-tools/kaleido-mind.mdx b/mintlify-docs/cn/ai-tools/kaleido-mind.mdx index 64fe52e..3aa0f41 100644 --- a/mintlify-docs/cn/ai-tools/kaleido-mind.mdx +++ b/mintlify-docs/cn/ai-tools/kaleido-mind.mdx @@ -20,7 +20,7 @@ description: "驱动 KaleidoSwap 代理式钱包的本地端推理引擎：分 |------|---------|------| | `T0` 快速通道 | "balance"、"address"、"btc price" | 零次推理，即时返回 | | `T2` recipe | "pay bob 3 EUR"、"buy 0.001 BTC" | 约一次推理（模型可协助提取参数槽），随后是确定性链路，需确认才继续 | -| `T1` 代理循环 | 其余全部请求 | 由 skill 限定范围的 LLM，可通过 P2P 把困难或新型链路委托给已配对桌面端的更大模型 | +| `T1` 代理循环 | 其余全部请求 | 由 skill 限定范围、运行在本地模型上的 LLM | - **T0，快速通道。** 确定性模式匹配，零推理。余额查询、地址、现货价格。 - **T2，recipe 引擎。** 由 skill 承载有序的执行计划（解析、定价、换算、确认、发送），模型只负责填充参数槽。这让多步流程即使在约 0.6B 参数的模型上也足够可靠，而不是要求模型自己规划整条链路。 @@ -32,7 +32,7 @@ description: "驱动 KaleidoSwap 代理式钱包的本地端推理引擎：分 | 接入方式 | 工具执行方式 | 花费前确认 | |---------|----------------|----------------------| -| 移动端（Rate） | 进程内 WDK 适配器（完全本地端、私密）；可选通过 P2P 委托给已配对的桌面端 | 确认面板 | +| 移动端（Rate） | 进程内 WDK 适配器（完全本地端、私密） | 确认面板 | | 桌面端 | 以 stdio 工具源方式接入 `kaleido-mcp`（命名空间为 `spark_*`/`rln_*`/`kaleidoswap_*`） | 确认对话框 | | 评测 / CLI | 预置的桩处理器，用于可复现的基准测试 | 自动批准（同时断言把关已触发） | @@ -61,11 +61,11 @@ Skills 是符合 Agent Skills 规范的操作手册（`SKILL.md` 加渐进式披 ## QVAC：本地端推理 -LLM、嵌入、语音转文字和文字转语音推理全部通过 Tether 的本地 AI 运行时 [QVAC SDK](https://www.npmjs.com/package/@qvac/sdk) 完成，默认在本地端运行，较重的任务也可以委托给用户明确配对并自行掌控的桌面端。它以 `@kaleidorg/mind/qvac` 子路径发布，使该 SDK 保持为 peer 依赖，而不是 core 的硬性依赖。记忆与 RAG（长期回忆、钱包历史检索、商户发现）同样走 QVAC 的嵌入能力，并通过近重复内容合并来避免记忆膨胀。 +LLM、嵌入、语音转文字和文字转语音推理全部通过 Tether 的本地 AI 运行时 [QVAC SDK](https://www.npmjs.com/package/@qvac/sdk) 完成，在本地设备上运行。将推理转移到其他设备的功能已在规划中，但目前尚不可用。它以 `@kaleidorg/mind/qvac` 子路径发布，使该 SDK 保持为 peer 依赖，而不是 core 的硬性依赖。记忆与 RAG（长期回忆、钱包历史检索、商户发现）同样走 QVAC 的嵌入能力，并通过近重复内容合并来避免记忆膨胀。 | 包 | 版本 | |---------|---------| -| `@qvac/sdk` | 核心引擎支持 0.13.1 及以上版本，并以 0.21 为测试基准。P2P 委托需要 0.13.1–0.18：QVAC 在 0.19 中移除了该功能 | +| `@qvac/sdk` | 核心引擎支持 0.13.1 及以上版本，并以 0.21 为测试基准 | | `@kaleidorg/mind` | 把 `@qvac/sdk` 作为可选 peer 依赖，因此引擎在完全没有模型的情况下也能配合 mock provider 运行 | ### 推荐模型 @@ -102,7 +102,7 @@ KaleidoMind 0.8 推荐 Qwen3.5 系列。下表中的每个模型都是 `@qvac/sd | 宿主 | 角色 | |------|------| | [Rate](https://github.com/kaleidoswap/Rate) | React Native 移动钱包，内含本地 LLM、语音转文字、神经网络文字转语音，以及免手操作的语音循环 | -| [桌面应用](/cn/desktop-app/getting-started/introduction) | 通过 Tauri sidecar（`apps/provider`）把引擎作为应用内聊天运行，该 sidecar 以 stdio 工具源方式接入 `kaleido-mcp`，同时还能作为手机的配对推理节点（桌面应用 v0.5.1 中手机配对功能暂停） | +| [桌面应用](/cn/desktop-app/getting-started/introduction) | 通过 Tauri sidecar（`apps/provider`）把引擎作为应用内聊天运行，该 sidecar 以 stdio 工具源方式接入 `kaleido-mcp` | 最快的体验方式是桌面应用。[下载最新版本](https://kaleidoswap.com/downloads)，然后按照[安装指南](/cn/desktop-app/getting-started/installation)操作。