mirocode 0.14.0 → 0.14.1

package/README.md

@@ -104,7 +104,6 @@ Type these in the interactive composer:
 | Command | Description |
 |---------|-------------|
 | `/init [extra]` | Analyze the codebase and create or improve `AGENTS.md` |
-| `/login` | Sign in to a subscription provider (ChatGPT/Codex, Claude, Copilot, OpenRouter, Kimi, or xAI) |
 | `/model [model]` | Select or switch model |
 | `/effort [level]` | Set the current model's effort level |
 | `/config [id] [val]` | View or change ACP session config options |
@@ -241,7 +240,7 @@ Use `/config sandbox on` or `/config sandbox off` to update this setting and swi
 
 Codemode is off by default and available only to the built-in Miro agent. Use `/config codemode on` to enable it or `/config codemode off` to disable it; both update `miro.codemode.enabled` in `settings.json`. Set it to `true` before starting a print-mode run. Normal tools remain available.
 
-The model can compose reads, searches, paging and aggregation in one JavaScript script inside an isolated QuickJS Worker, returning only selected output to its context. Scripts can call `read_file`, `grep`, `glob` and configured MCP bridges; disabled tools stay unavailable. They have no host `process`, `fetch`, file writes, Terminal or subagent access. Each actual MCP call still follows the normal permission policy, including denial in non-interactive Manual mode, and internal calls remain visible in the transcript.
+The model can compose reads, searches, paging and aggregation in one JavaScript script inside an isolated QuickJS Worker, returning only selected output to its context. Scripts can call `read_file`, `grep`, `glob` and configured MCP bridges. They have no host `process`, `fetch`, file writes, Terminal or subagent access. Each actual MCP call still follows the normal permission policy, including denial in non-interactive Manual mode, and internal calls remain visible in the transcript.
 
 Each script has a 60-second VM limit, a 64 MiB QuickJS heap, at most 64 tool calls and eight concurrent reads, with separate tool-data and output budgets. Scripts must await their calls. Cancellation and timeout wait for host cleanup but cannot undo remote effects; failed scripts are never automatically replayed. See the [tool reference](./docs/tools/codemode.md) for the exact API and limits. This is separate from the Terminal sandbox and does not change ACP providers.
 
@@ -311,7 +310,7 @@ Composer history is independent of the transcript, at `~/.miro/history.jsonl`.
 
 ## Configuration
 
-Model, effort, prompt-language, status-line, and custom ACP-provider preferences live in `~/.miro/settings.json`. Its legacy top-level `provider` value is ignored. Miro LLM endpoints and models live in `~/.miro/models.json`; subscription OAuth credentials created by `/login` live separately in `~/.miro/auth.json`. `/login` is available in the Miro TUI and opens the provider's browser or device-code flow. OAuth subscriptions and API keys are separate authentication paths; keep `auth.json` private and use subscription credentials only for your own local work. `/config` changes both the live ACP options advertised as `configOptions` and miro's own `language` option.
+Model, effort, prompt-language, status-line, and custom ACP-provider preferences live in `~/.miro/settings.json`. Its legacy top-level `provider` value is ignored. Miro LLM endpoints and models are imported only from `~/.miro/models.json`. `/config` changes both the live ACP options advertised as `configOptions` and miro's own `language` option.
 
 ### Permission mode
 

package/README.zh.md

@@ -22,7 +22,7 @@
 ## 环境要求
 
 - [Bun](https://bun.sh/) **>= 1.4.3**；装 npm 包时只需 Node **>= 22.7**
-- 为 Miro 自带 agent 准备一个兼容的 API 端点；仅在使用 `--acp` 时才需要 `PATH` 上的 ACP 提供方二进制（见 [ACP 提供方](#acp-提供方）)
+- 为 Miro 自带 agent 准备一个兼容的 API 端点；仅在使用 `--acp` 时才需要 `PATH` 上的 ACP 提供方二进制（见 [ACP 提供方](#acp-提供方)）
 
 ## 安装与运行
 
@@ -104,7 +104,6 @@ bun start -- -p "Analyze this project"
 | 命令 | 说明 |
 |------|------|
 | `/init [extra]` | 分析代码库并创建或改进 `AGENTS.md` |
-| `/login` | 登录订阅提供方（ChatGPT/Codex、Claude、Copilot、OpenRouter、Kimi 或 xAI） |
 | `/model [model]` | 选择或切换模型 |
 | `/effort [level]` | 设置当前模型的努力程度 |
 | `/config [id] [val]` | 查看或修改 ACP 会话配置项 |
@@ -241,7 +240,7 @@ miro 用到的上游写在 `~/.miro/models.json`。打开 `/model` 会重新读
 
 Codemode 默认关闭，仅适用于内置 Miro agent。用 `/config codemode on` 开启、`/config codemode off` 关闭，两者都会修改 `settings.json` 的 `miro.codemode.enabled`。Print 模式需在启动前将它设为 `true`。普通工具仍然保留。
 
-模型可在独立 QuickJS Worker 内用一段 JavaScript 组合读取、搜索、分页和聚合，只把选定输出交回上下文。脚本能调用 `read_file`、`grep`、`glob` 和已配置的 MCP 桥接，禁用工具不会重新开放；没有宿主 `process`、`fetch`、写文件、Terminal 或 subagent 能力。每次实际 MCP 调用仍遵循普通权限策略，非交互 Manual 会拒绝调用，内部调用保留在 transcript 中。
+模型可在独立 QuickJS Worker 内用一段 JavaScript 组合读取、搜索、分页和聚合，只把选定输出交回上下文。脚本能调用 `read_file`、`grep`、`glob` 和已配置的 MCP 桥接；没有宿主 `process`、`fetch`、写文件、Terminal 或 subagent 能力。每次实际 MCP 调用仍遵循普通权限策略，非交互 Manual 会拒绝调用，内部调用保留在 transcript 中。
 
 每段脚本的 VM 时限为 60 秒、QuickJS 堆为 64 MiB，最多调用 64 次工具、并发读取 8 条，另有工具数据和输出预算。脚本必须 await 工具调用。取消与超时会等待宿主收尾，但不能撤销远端副作用；失败脚本不会自动重放。精确 API 和限制见[工具参考](./docs/tools/codemode.md)。它与 Terminal 沙箱相互独立，不改变 ACP 提供方行为。
 
@@ -311,7 +310,7 @@ miro 会在可见聊天记录之外保存模型上下文检查点，包含压缩
 
 ## 配置
 
-模型、努力程度、提示词语言、状态栏和自定义 ACP 提供方偏好保存在 `~/.miro/settings.json`；其中遗留的顶层 `provider` 会被忽略。miro 的上游与模型列表保存在 `~/.miro/models.json`；`/login` 创建的订阅 OAuth 凭据独立保存在 `~/.miro/auth.json`。`/login` 仅在 miro TUI 中可用，会启动提供方的浏览器或设备码流程。OAuth 订阅与 API key 是独立认证路径；请妥善保管 `auth.json`，订阅凭据只应用于你自己的本地使用。`/config` 既可修改 ACP 以 `configOptions` 声明的实时选项，也可修改 miro 自己的 `language` 选项。
+模型、努力程度、提示词语言、状态栏和自定义 ACP 提供方偏好保存在 `~/.miro/settings.json`；其中遗留的顶层 `provider` 会被忽略。miro 的上游与模型列表只从 `~/.miro/models.json` 导入。`/config` 既可修改 ACP 以 `configOptions` 声明的实时选项，也可修改 miro 自己的 `language` 选项。
 
 ### 权限模式