Baize MCP 0.1.4

Baize MCP v0.1.4

中文

Baize MCP v0.1.4 在命令计划工作流之外补齐了观察、诊断与远程任务闭环，让 AI 客户端可以在白泽账号权限范围内完成"了解状态 → 定位问题 → 执行/跟踪任务"的完整流程，同时保持结果字段、数量和文本长度的有界输出。

主要变化

• 增加运行概览工具，读取当前账号可见范围内的平台运行摘要和有限数量的重点异常节点；缓存缺失、异常列表为空、分区失败均显式标记。
• 增加节点有界观察工具，按健康、指标、进程、存储、Docker、Nginx、主机画像状态和控制面等维度读取单一节点；敏感正文、凭据、环境变量和完整历史不返回。
• 增加固定语义观察工具：告警列表、资产清单与到期、证书监控状态、定时任务、日志、Nginx、Runbook、安全暴露面、订阅状态和组件版本状态；各视图只返回完成判断所需字段，关联标识、地址和原始路径均已脱敏。
• 增加只读运行态诊断：可以为单个节点发起诊断探测、查询状态与证据数量，并在脱敏后读取诊断 AI 上下文。
• 补齐远程任务闭环：可以通过直接任务入口创建模板任务或服务端允许的精确自定义命令，把等待中的任务推入执行链路，在用户明确要求后按目标与游标读取有界输出，并继续支持查询进度和请求取消。
• 增加告警确认/解决工具；服务端负责权限、状态规则和审计，成功后需再次查询确认最终状态。
• 增加工作流状态工具：读取本机单人/多人工作流模式和服务端审批策略摘要；无策略查看权限时仍返回本地模式并显式标记。
• 增加命令计划取消工具；错误信息保留稳定的原因、可重试标记、消息键和下一步动作键，不再透传底层错误原文。
• AI 接入说明覆盖 DeepSeek Harness（DSH）：接入安装器会自动注册，也给出用户插件层的标准配置。
• 最低兼容白泽版本提升为 0.2.2，以使用直接任务派发等新发布的稳定行为。

兼容性与升级

• 最低兼容白泽版本：0.2.2。
• 传输方式：stdio。
• 支持 Linux、macOS、Windows 的 amd64 与 arm64 发布包，每个包内含可执行文件、启动校验文件、项目许可证、中英文说明和第三方许可证清单。
• 已安装用户可从此前获取的 baize 公开入口运行 AI 接入升级器；升级会校验发布归档和可执行文件，并保留本机 profile 与系统凭据。升级前退出正在使用 MCP 的 AI 客户端，完成后重新打开。
• 手动下载时请在 GitHub Releases 核对 SHA256SUMS；每个发布包内含 baize-mcp.sha256 完整性文件，程序启动时会自动校验可执行文件，发现文件损坏或安装不完整即拒绝运行。
• 白泽权限、审批策略、风险确认和审计继续由服务端处理；MCP 不复制这些控制，审批通过也不代表计划会自动执行。

完整安装和使用说明见 Baize 公开入口 的 AI 接入与远程任务指南。

English

Baize MCP v0.1.4 completes the loop beyond the command-plan workflow: observation, diagnosis, and remote-task execution, so an AI client can cover "understand state → locate issues → execute and track tasks" within the signed-in Baize account's permissions while keeping bounded fields, counts, and text lengths.

Highlights

• Adds a runtime overview tool with the account-scoped platform summary and a bounded set of highlighted abnormal nodes; missing caches, an empty abnormal list, and failed sections are marked explicitly.
• Adds bounded agent observation for one node across health, metrics, processes, storage, Docker, Nginx, host-profile, and control-plane views; sensitive bodies, credentials, environment values, and complete histories are excluded.
• Adds fixed semantic observation tools: alert lists, asset inventory and expiry, certificate status, scheduled tasks, logs, Nginx, Runbooks, security exposure, subscription status, and component version status; each view returns only the fields needed for the current decision, with correlation identifiers, addresses, and raw paths redacted.
• Adds read-only runtime diagnosis: start a probe for one node, query its status and evidence counts, and read the privacy-reduced diagnosis AI context.
• Closes the remote-task loop: create template tasks or exact custom commands allowed by Baize through the direct-task entry, push pending tasks into the execution chain, read bounded output by target and cursor after the user explicitly asks for it, and keep progress lookup and cancellation.
• Adds alert acknowledgement/resolution; Baize handles permissions, state rules, and audit, and callers must query the alert again to confirm the final status.
• Adds a workflow-status tool reading the local single-user/multi-user mode and a minimal server approval-policy summary; when policies are not viewable, the local mode is still returned with an explicit marker.
• Adds command-plan cancellation, and keeps stable reason, retryable, message-key, and next-action fields in errors instead of exposing raw backend messages.
• AI access documentation now covers DeepSeek Harness (DSH): the installer registers it automatically, and the standard patch-layer configuration is documented.
• The minimum compatible Baize version is raised to 0.2.2 for the newly released stable direct-task dispatch and related behavior.

Compatibility and upgrade

• Minimum compatible Baize version: 0.2.2.
• Transport: stdio.
• Release archives are available for Linux, macOS, and Windows on amd64 and arm64; each contains the executable, startup integrity file, project license, bilingual guides, and third-party license notices.
• Existing users can run the AI access upgrader from the previously obtained baize public entry. The upgrader verifies the release archive and executable, and preserves the local profile and operating-system credential store. Close AI clients using MCP before upgrading, then reopen them afterward.
• For manual downloads, verify the archive against SHA256SUMS on GitHub Releases; every archive also ships a baize-mcp.sha256 integrity file and the executable verifies itself at startup, refusing to run when the file is corrupted or the installation is incomplete.
• Baize permissions, approval policy, risk confirmation, and audit remain server-side; MCP does not replicate those controls, and approval does not cause a plan to execute automatically.

See the Baize public entry and its AI Access and Remote Task Guide for complete installation and usage instructions.

Baize MCP 0.1.3

Baize MCP v0.1.3

中文

Baize MCP v0.1.3 增加命令计划审批闭环，让 AI 客户端可以在白泽现有账号权限和审批规则范围内申请、查看并提交审批决定，同时继续保持执行前的独立确认。

主要变化

• 增加命令计划审批申请、分页列表、详情和通过/驳回决策工具。
• 审批通过只提交当前账号的后端决策；权限、自审批限制、计划快照、过期时间和审计仍由白泽处理，不会因为 MCP 自动获得权限，也不会自动执行计划。
• 增加分页继续标记，并为工具目录和结构化结果设置 64 KiB 上限，减少异常响应占用 AI 上下文。
• 改进跨平台 AI 接入升级流程，覆盖 Linux、macOS 和 Windows。
• 源码构建最低版本更新为 Go 1.25.13，以使用当前标准库安全修复。

兼容性与升级

• 最低兼容白泽版本：0.2.1。
• 传输方式：stdio。
• 支持 Linux、macOS、Windows 的 amd64 与 arm64 发布包。
• 已安装用户可从此前获取的 baize 公开入口运行 AI 接入升级器。升级会校验发布归档和可执行文件，并保留本机 profile 与系统凭据；升级前退出正在使用 MCP 的 AI 客户端，完成后重新打开客户端。
• 审批通过后仍需单独确认执行；白泽要求的风险确认、二次确认或其它审批不得绕过。

完整安装和使用说明见 Baize 公开入口 的 AI 接入与远程任务指南。

English

Baize MCP v0.1.3 adds a command-plan approval workflow. AI clients can request, inspect, and submit approval decisions within the signed-in Baize account's existing permissions and approval rules, while execution remains a separate confirmation.

Highlights

• Added command-plan approval request, paginated listing, detail, and approve/reject decision tools.
• Approval decisions are submitted to Baize for backend permission checks. MCP does not grant permission locally, bypass self-approval restrictions, ignore snapshot or expiry checks, or execute a plan automatically.
• Added pagination continuation markers and 64 KiB limits for the tool catalog and structured results to keep abnormal responses from consuming excessive AI context.
• Improved cross-platform AI access upgrades across Linux, macOS, and Windows.
• Raised the minimum Go version for source builds to Go 1.25.13 so builds use the current standard-library security fixes.

Compatibility and upgrade

• Minimum compatible Baize version: 0.2.1.
• Transport: stdio.
• Release archives are available for Linux, macOS, and Windows on amd64 and arm64.
• Existing users can run the AI access upgrader from the previously obtained baize public entry. The upgrader verifies the release archive and executable, and preserves the local profile and operating-system credential store. Close AI clients using MCP before upgrading, then reopen them afterward.
• Approval does not execute a plan. Keep separate confirmation for execution and follow any risk, secondary, or additional approval requested by Baize.

See the Baize public entry and its AI Access and Remote Task Guide for complete installation and usage instructions.

Baize MCP 0.1.2

Baize MCP v0.1.2

中文

Baize MCP v0.1.2 增加了第一批命令工作流工具。AI 客户端现在可以在本机安全会话的权限范围内发现白泽命令模板、预览模板执行条件、创建命令计划，并跟踪或请求取消远程任务。

主要变化

• 增加 baize_command_templates_list，查询当前账号可用的模板摘要和参数约束。
• 增加 baize_command_template_preview，对指定节点进行渲染预览和预检；预览不会创建计划或运行命令。
• 增加 baize_command_plan_create、baize_command_plan_get 和 baize_command_plan_execute，创建计划不会直接派发命令。
• 增加 baize_exec_task_get 和 baize_exec_task_cancel，查看任务进度或请求取消等待中、运行中的任务。
• 写工具继续使用白泽现有账号权限、风险确认、审批、审计和任务状态控制，MCP 不建立第二套权限或审批逻辑。
• 计划和任务结果采用有限字段、数量和文本长度边界，不返回命令正文、工作目录、环境变量、操作者身份、任务输出或凭据。

兼容性与升级

• 最低兼容白泽版本：0.2.1。
• 传输方式：stdio。
• 已安装用户可从此前获取的 baize 公开入口运行 AI 接入升级器。升级会校验发布归档和可执行文件，并保留本机 profile 与系统凭据；升级前退出正在使用 MCP 的 AI 客户端，完成后重新打开客户端。
• 创建计划不会自动执行。执行或取消任务前，AI 应展示目标、目的、风险、预期结果和回滚方式，并等待操作者明确确认；白泽返回需要审批或二次确认时不得绕过。

完整安装和使用说明见 Baize 公开入口 的 AI 接入与远程任务指南。

English

Baize MCP v0.1.2 adds the first command workflow tools. Within the permissions of the local Baize session, an AI client can discover command templates, preview execution conditions, create command plans, and track or request cancellation of remote tasks.

Highlights

• Added baize_command_templates_list for template summaries and parameter constraints allowed for the signed-in account.
• Added baize_command_template_preview for rendering previews and prechecks on selected agents; preview never creates a plan or runs a command.
• Added baize_command_plan_create, baize_command_plan_get, and baize_command_plan_execute; creating a plan does not dispatch a command.
• Added baize_exec_task_get and baize_exec_task_cancel to inspect progress or request cancellation of pending or running tasks.
• Write tools continue to use Baize's existing account permissions, risk confirmation, approval, audit, and task-state controls. MCP does not add a second permission or approval layer.
• Plan and task results use bounded fields, item counts, and text lengths, and exclude command bodies, working directories, environment values, operator identity, task output, and credentials.

Compatibility and upgrade

• Minimum compatible Baize version: 0.2.1.
• Transport: stdio.
• Existing users can run the AI access upgrader from the previously obtained baize public entry. The upgrader verifies the release archive and executable, and preserves the local profile and operating-system credential store. Close AI clients using MCP before upgrading, then reopen them afterward.
• Creating a plan does not execute it. Before execution or cancellation, the AI should show the target, purpose, risk, expected result, and rollback path and wait for explicit operator confirmation. Never bypass an approval or secondary confirmation requested by Baize.

See the Baize public entry and its AI Access and Remote Task Guide for complete installation and usage instructions.

Baize MCP 0.1.1

Baize MCP 0.1.1

中文

本版本增强了 AI 定位白泽服务器节点时的筛选和排序能力，适用于节点较多或需要按环境快速查找节点的场景。

更新内容

• baize_agents_list 新增按别名、系统、地区、架构、Agent 版本和分组筛选。
• 支持按创建时间、更新时间、最后心跳、注册时间、名称、状态、版本或系统排序。
• 仅暴露已经验证可用的筛选条件。
• 保持现有三个工具名称、登录方式、客户端配置和隐私保护结果兼容。
• 当前版本仍为只读工具，不改变白泽账号已有的权限范围。

安装与兼容性

• 提供 Linux、macOS 和 Windows 的 amd64、arm64 运行包。
• 发布包启动时会自动检查可执行文件完整性。
• 升级后请重启 AI 客户端或 MCP 进程，使新的工具定义生效。
• 最低兼容版本：Baize 0.2.1。

English

This release improves how AI clients locate Baize server agents, especially when working with larger environments.

Changes

• baize_agents_list can now filter by alias, system, region, architecture, Agent version, and group.
• Results can be sorted by creation time, update time, last heartbeat, registration time, name, status, version, or operating system.
• Only verified and usable filters are exposed.
• Existing tool names, sign-in flow, client configuration, and privacy-protected outputs remain compatible.
• All tools in this release remain read-only and use the permissions of the signed-in Baize account.

Installation and Compatibility

• Runtime archives are provided for Linux, macOS, and Windows on amd64 and arm64.
• Each archive performs an automatic executable integrity check at startup.
• Restart the AI client or MCP process after upgrading so the updated tool definitions are loaded.
• Minimum supported Baize version: 0.2.1.

Baize MCP 0.1.0

Baize MCP 0.1.0

中文

Baize MCP 的首个可运行版本，让支持 MCP 的 AI 软件、IDE 和终端客户端通过 stdio 连接你自己的白泽实例，并在现有账号权限范围内读取经过隐私保护的服务器节点状态信息。

新增

• 提供 baize_connection_status、baize_agents_list 和 baize_agent_get 三个只读工具，用于检查连接、分页查询节点以及读取单个节点的基础状态。
• 提供本机交互式登录、会话检查、退出登录和多 profile 支持；AI 客户端配置只需保存可执行文件路径和 profile 名称。
• 提供 Linux、macOS、Windows 的 amd64 与 arm64 发布包。

安全边界

• 密码只从本机交互式终端读取，不进入命令参数或普通配置文件；用户名不会持久化，会话凭据由操作系统凭据存储保护。
• 工具结果只返回节点 ID、显示名称、状态、操作系统、架构、Agent 版本和最后心跳时间，不返回连接地址、IP、指纹、能力清单、凭据或服务端追踪信息。
• 当前工具均为只读、幂等、非破坏性工具；底层错误会转换为有限且不包含连接细节的提示。
• 默认要求 HTTPS；本机回环地址可使用 HTTP，其它 HTTP 地址需要用户显式允许。
• 发布包启动时会自动校验可执行文件 SHA-256；校验文件缺失或不匹配时不会继续运行。

安装与兼容性

• 最低兼容版本为 Baize 0.2.1。
• 请从本 Release 下载与你的系统和架构匹配的压缩包并完整解压；程序会自动执行启动自检，SHA256SUMS 仍可用于额外的发布文件校验。
• 每个压缩包包含可执行文件、启动校验文件、版本文件、Apache-2.0 许可证、中英文说明和实际依赖的第三方许可证文本。

未来方向

当前版本只提供只读能力。后续计划在完成权限范围、明确确认、风险提示、审计和可回滚约束后，逐步增加受控写操作；具体能力以正式版本说明为准。

English

Baize MCP 0.1.0 is the first runnable release of the open-source MCP connector for Baize. It lets MCP-compatible AI applications, IDEs, and terminal clients connect to a user-managed Baize instance over stdio and read privacy-protected server status information within the signed-in user's existing permissions.

Added

• Added three read-only tools: baize_connection_status, baize_agents_list, and baize_agent_get for connection checks, paginated agent queries, and basic status for one agent.
• Added interactive local sign-in, session checks, sign-out, and multiple profiles. AI client configuration only needs the executable path and profile name.
• Added amd64 and arm64 release archives for Linux, macOS, and Windows.

Security Boundary

• Passwords are read only from an interactive local terminal and are never placed in command arguments or regular configuration files. Usernames are not persisted, and session credentials are protected by the operating system's credential store.
• Tool results contain only the agent ID, display name, status, operating system, architecture, Agent version, and last heartbeat. Connection addresses, IP addresses, fingerprints, capability lists, credentials, and server trace information are excluded.
• All current tools are read-only, idempotent, and non-destructive. Lower-level errors are converted into limited messages without connection details.
• HTTPS is required by default. HTTP is accepted for loopback addresses; any other HTTP address requires explicit user approval.
• Release archives automatically verify the executable SHA-256 at startup and refuse to run when the integrity file is missing or does not match.

Installation and Compatibility

• The minimum compatible Baize version is 0.2.1.
• Download and fully extract the archive for your operating system and architecture from this Release. Startup performs the integrity check automatically, while SHA256SUMS remains available for optional release-file verification.
• Every archive includes the executable, startup integrity file, version file, Apache-2.0 license, Chinese and English guides, and the license texts for the dependencies included in the build.

Future Direction

This version is read-only. Future releases may add controlled write operations after permission scope, explicit confirmation, risk prompts, auditability, and rollback constraints are in place; formal release notes will define the actual scope.