@luffysolution/omnischolar-pi 0.1.2 → 0.1.3

package/README.md

@@ -60,7 +60,7 @@ In the ChatGPT desktop app, restart the app, open **Plugins**, select the **Omni
 The plugin bundles all eight Skills and starts its local MCP server with the pinned PyPI release:
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 No separate `pip install` is required for this plugin path. The first MCP start needs network access so `uvx` can download and cache the package. Provider credentials and optional service settings remain in your OmniScholar configuration; plugin installation does not collect them.
@@ -100,7 +100,7 @@ omnischolar update cursor --scope project
 omnischolar uninstall cursor --scope project
 ```
 
-For Pi, the full installer runs `pi install npm:@luffysolution/omnischolar-pi` and installs the bundled Skills separately. The npm Extension starts `omnischolar mcp`, discovers its tools, and registers them with Pi. You can also install the Extension directly:
+For Pi, the full installer runs `pi install npm:@luffysolution/omnischolar-pi` and installs the bundled Skills separately. The npm Extension starts the matching PyPI MCP through `uvx`, discovers its tools, and registers them with Pi; a separate global Python CLI installation is not required. You can also install the Extension directly:
 
 ```sh
 pi install npm:@luffysolution/omnischolar-pi
@@ -134,7 +134,7 @@ Create a labelled illustration of this mechanism. Treat it as a draft, not exper
 
 ## Configuration
 
-Copy [`omnischolar.config.example.json`](omnischolar.config.example.json) to `omnischolar.config.json`. Keep API keys in environment variables and refer to their names with `apiKeyEnv`.
+On the first MCP start or installer run, if no discoverable config exists, OmniScholar creates one user-level `omnischolar.config.json`. You can also create it explicitly with `omnischolar config init`. Keep API keys in environment variables and refer to their names with `apiKeyEnv`.
 
 A small local configuration can start with Zotero and the output directory:
 

package/README.zh-CN.md

@@ -58,7 +58,7 @@ codex plugin add omnischolar@omnischolar
 插件会同时安装 8 个 Skills，并使用固定的 PyPI 版本启动本地 MCP：
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 采用插件安装方式时无需另外执行 `pip install`。第一次启动 MCP 需要联网，以便 `uvx` 下载并缓存包。各服务的凭据与可选配置仍保存在 OmniScholar 配置中，插件安装不会收集这些信息。
@@ -98,7 +98,7 @@ omnischolar update cursor --scope project
 omnischolar uninstall cursor --scope project
 ```
 
-对 Pi，完整安装器会运行 `pi install npm:@luffysolution/omnischolar-pi`，并单独安装随包提供的 Skills。npm Extension 会启动 `omnischolar mcp`、发现工具并注册到 Pi。也可以直接安装 Extension：
+对 Pi，完整安装器会运行 `pi install npm:@luffysolution/omnischolar-pi`，并单独安装随包提供的 Skills。npm Extension 会通过 `uvx` 启动与当前版本匹配的 PyPI MCP、发现工具并注册到 Pi，因此无需另行安装全局 Python 命令。也可以直接安装 Extension：
 
 ```sh
 pi install npm:@luffysolution/omnischolar-pi
@@ -132,7 +132,7 @@ omnischolar mcp
 
 ## 配置
 
-将 [`omnischolar.config.example.json`](omnischolar.config.example.json) 复制为 `omnischolar.config.json`。API key 建议放在环境变量中，配置文件只填写变量名 `apiKeyEnv`。
+首次启动 MCP 或执行安装命令时，如果没有可发现的配置，程序会在用户配置目录自动创建唯一的 `omnischolar.config.json`。也可以用 `omnischolar config init` 主动创建。API key 建议放在环境变量中，配置文件只填写变量名 `apiKeyEnv`。
 
 最小的本地配置可以只写 Zotero 和输出目录：
 

package/docs/CONFIGURATION.en.md

@@ -16,6 +16,8 @@ The first matching file is used; files are not merged:
 
 Relative paths resolve from the config file. Misspelled fields and invalid values are rejected.
 
+If none of the first four sources exists, the first MCP start or installer run creates one user-level configuration file. On Windows this is normally `%LOCALAPPDATA%\\omnischolar\\omnischolar.config.json`; other platforms follow `platformdirs`. You can also run `omnischolar config init`. Existing configuration is never overwritten.
+
 ## Minimal config
 
 ```json

package/docs/CONFIGURATION.md

@@ -16,6 +16,8 @@ OmniScholar 使用 `schemaVersion: 1` 的 JSON 配置。可直接复制根目录
 
 相对路径以配置文件所在目录为基准。拼错字段名或填写不合法的值时，程序会直接报错。
 
+如果前四项都不存在，第一次启动 MCP 或执行安装命令时会自动创建用户级配置文件。Windows 通常位于 `%LOCALAPPDATA%\omnischolar\omnischolar.config.json`，其他平台由 `platformdirs` 决定。也可以运行 `omnischolar config init` 主动创建；程序不会覆盖已有配置。
+
 ## 最小配置
 
 ```json

package/docs/INSTALLATION.en.md

@@ -44,7 +44,7 @@ Restart the ChatGPT desktop app after adding the marketplace. Open **Plugins**,
 The installed plugin contains `plugin.json`, `skills/`, and `mcp.json`. Its MCP entry executes:
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 `uvx` creates an isolated environment and caches the exact release, so this route does not require a separate `pip install`. The first start requires package-index access. To fetch updates to the Git marketplace, run:
@@ -83,7 +83,7 @@ Choose `--scope project` to share the enabled plugin through repository settings
 Claude Code copies the repository-root plugin into its versioned cache, discovers the eight folders under `skills/`, and starts `.mcp.json` automatically when the plugin is enabled. The MCP command is pinned to:
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 The first MCP start needs package-index access. Skills are namespaced with `omnischolar`, for example `/omnischolar:scholar-search` and `/omnischolar:zotero-research`.
@@ -121,7 +121,7 @@ omnischolar install cursor --scope project
 | Pi | npm Extension | npm Extension | supported |
 | WorkBuddy/CodeBuddy | `~/.codebuddy/.mcp.json` | `.mcp.json` | no documented portable path |
 
-For Pi, `omnischolar install pi` installs `npm:@luffysolution/omnischolar-pi` with Pi's package manager and copies the Skills to the selected scope. The Extension starts the local `omnischolar mcp` process. The `pi` command and the Python `omnischolar` command must both be on `PATH`.
+For Pi, `omnischolar install pi` installs `npm:@luffysolution/omnischolar-pi` with Pi's package manager and copies the Skills to the selected scope. The Extension starts the matching PyPI MCP through `uvx`, so only `uv` must be on `PATH`; a separate global Python CLI installation is not required.
 
 WorkBuddy/CodeBuddy supports local stdio MCP servers. Its official documentation does not define a portable Skills directory, so only the MCP component is installed automatically.
 

package/docs/INSTALLATION.md

@@ -44,7 +44,7 @@ codex plugin add omnischolar@omnischolar
 安装后的插件包含 `plugin.json`、`skills/` 和 `mcp.json`。其 MCP 配置执行：
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 `uvx` 会创建隔离环境并缓存该精确版本，因此无需另行运行 `pip install`。第一次启动需要能够访问 Python 包索引。若要更新 Git Marketplace：
@@ -83,7 +83,7 @@ claude plugin install omnischolar@omnischolar --scope user
 Claude Code 会把仓库根插件复制到版本化缓存，发现 `skills/` 下的 8 个 Skills，并在插件启用时自动启动 `.mcp.json`。MCP 命令固定为：
 
 ```sh
-uvx --from luffysolution-omnischolar==0.1.2 omnischolar mcp
+uvx --from luffysolution-omnischolar==0.1.3 omnischolar mcp
 ```
 
 第一次启动 MCP 需要能够访问 Python 包索引。Skills 使用 `omnischolar` 命名空间，例如 `/omnischolar:scholar-search` 和 `/omnischolar:zotero-research`。
@@ -121,7 +121,7 @@ omnischolar install cursor --scope project
 | Pi | npm Extension | npm Extension | 支持 |
 | WorkBuddy/CodeBuddy | `~/.codebuddy/.mcp.json` | `.mcp.json` | 无官方可移植目录 |
 
-对 Pi，`omnischolar install pi` 会通过 Pi 包管理器安装 `npm:@luffysolution/omnischolar-pi`，并把 Skills 复制到所选作用域。Extension 会启动本地 `omnischolar mcp` 进程，因此 `pi` 与 Python 的 `omnischolar` 命令都必须位于 `PATH`。
+对 Pi，`omnischolar install pi` 会通过 Pi 包管理器安装 `npm:@luffysolution/omnischolar-pi`，并把 Skills 复制到所选作用域。Extension 会通过 `uvx` 启动与当前版本匹配的 PyPI MCP，因此只需保证 `uv` 位于 `PATH`，无需另行安装全局 Python 命令。
 
 WorkBuddy/CodeBuddy 支持本地 stdio MCP。其官方文档没有定义可移植的 Skills 目录，因此安装器只自动配置 MCP。
 

package/host-configs/pi.md

@@ -8,6 +8,6 @@ omnischolar install pi
 pi install npm:@luffysolution/omnischolar-pi
 ```
 
-The Extension starts `omnischolar mcp`, discovers the server's tools, forwards calls and cancellation signals, and closes the MCP client when the Pi session ends.
+The Extension starts the matching PyPI MCP through `uvx`, discovers the server's tools, forwards calls and cancellation signals, and closes the MCP client when the Pi session ends. A separate global `omnischolar` command is not required; `uv` must be on `PATH`.
 
 `omnischolar mcp install pi` remains `manual_required` because Pi uses an Extension instead of a native MCP settings file. Use the full `omnischolar install pi` command.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@luffysolution/omnischolar-pi",
-  "version": "0.1.2",
+  "version": "0.1.3",
   "description": "Pi MCP bridge for the OmniScholar research toolkit",
   "type": "module",
   "license": "MIT",

package/pi-extension/dist/index.js

@@ -1,7 +1,7 @@
 import { Client } from "@modelcontextprotocol/sdk/client/index.js";
 import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
 import { Type } from "typebox";
-const CLIENT_INFO = { name: "omnischolar-pi", version: "0.1.2" };
+const CLIENT_INFO = { name: "omnischolar-pi", version: "0.1.3" };
 export function resultText(result) {
     return result.content
         .map((item) => {
@@ -38,8 +38,8 @@ export default function omnischolarExtension(pi) {
             return connection;
         connection = (async () => {
             transport = new StdioClientTransport({
-                command: "omnischolar",
-                args: ["mcp"],
+                command: "uvx",
+                args: ["--from", "luffysolution-omnischolar==0.1.3", "omnischolar", "mcp"],
                 stderr: "pipe",
             });
             client = new Client(CLIENT_INFO);