GenericAgent Admin Go

Desktop Management Panel for GenericAgent

Go backend + React frontend · Single executable · Cross-platform

📌 Requires: This admin panel manages local GenericAgent instances. Install GenericAgent first.

🌟 Overview

GenericAgent Admin Go is a desktop management panel for GenericAgent. A Go backend handles processes, files, configuration, updates, and system integration, while a React/Vite frontend provides a control console. It packages into a single ga-admin / ga-admin.exe executable.

The goal is not to replace GenericAgent, but to consolidate local GA runtime state, task entry points, model configuration, team collaboration, and desktop assistance into a maintainable UI.

🖥️ See It in Action

Native chat with instance switching, model selection, streaming replies, and usage tracking.

📋 Key Features

Feature	Description
📊 Dashboard	Overview of service status, recent activity, system stats
💬 Native Chat	Chat interface with `/chat` entrypoint, streaming response, usage tracking, model switching
📝 File Editor	Browse GA root, edit skills/SOPs/configs, syntax highlighting, file tree
📋 Task Management	Services (start/stop worker), scheduled tasks, Goal runs, autonomous reports
🧠 Memory Browser	View/search global and project memory
📡 Channel Monitor	Active channels, message logs
🤖 Autonomous Mode	Background task execution
📈 Usage Tracking	Token/cost statistics per model
🎯 Goal Mode	Start/stop long-running goals, view logs/status
⚙️ Model Config	Wizard to add models, test endpoints, manage profiles
🔧 Settings	App config, auth, ga_root path
📄 Log Viewer	Tail worker logs, search history

📈 Star History

🚀 Quick Start

⚠️ Prerequisites: Python 3.11+ (for GenericAgent), Node.js 18+ / Go 1.22+ (for development builds)

For LLM Agents

Fetch installation guide and follow:

curl -fsSL https://raw.githubusercontent.com/Fwind43/GenericAgent-Admin/main/README.md

For Human Users

Method 1 — Download Release Package (Recommended)

Download the platform-specific package from GitHub Releases:

ga-admin-windows-amd64.zip
ga-admin-linux-amd64.tar.gz
ga-admin-darwin-amd64.tar.gz  (macOS Intel)
ga-admin-darwin-arm64.tar.gz  (macOS Apple Silicon)

Extract and create config.local.json in the same directory:

{
  "ga_root": "E:/Work/GenericAgent"
}

Windows: Double-click ga-admin.exe. The UI opens in a native desktop window backed by the WebView2 runtime (preinstalled on Windows 11 and current Windows 10), and the app keeps running in the system tray after you close the window. Run ga-admin.exe --no-window to use your default browser instead, or ga-admin.exe --no-browser to start without opening any UI. macOS: Run ./ga-admin. The UI opens in a native desktop window backed by WKWebView (built into macOS). Closing the window keeps the app in the menu bar. Use --no-window to open your default browser instead, or --no-browser to start without opening any UI. Linux: Run ./ga-admin or ./ga-admin --no-browser. Linux still opens the UI in your default browser.

By default the server listens on 127.0.0.1 with a random port, so nothing is exposed to the network and local access needs no password. The address of the running process is printed at startup and written to runtime.local.json; open that URL if you want a second view of the UI. To reach the admin server from another device, turn on remote access in Settings (see Remote access).

Method 2 — Local Development Build

cd GenericAgent-Admin-Go
npm --prefix web install
npm --prefix web run build
go run .

Open the URL printed on startup, or run go run . --port 8787 to pin a loopback port. The Vite dev server (npm --prefix web run dev) proxies /api to the port recorded in runtime.local.json, so it follows the random port automatically.

🎯 Core Capabilities

For Daily Users

Service Management: Start/stop worker, monitor logs, check process status
Chat Interface: /chat entrypoint with streaming, usage tracking, model switching

Choosing a Project Mode

Open Chat settings → Project mode → Global runtime mode, choose a mode, and click Save. This changes project memory handling, not the model or a capability tier.

Mode	Memory organization	When to choose it
Official project mode	Uses GenericAgent's official project mechanism and the project's `project_memory.md`	Keep the official workflow
Admin project mode (L1–L3)	L1 persistent rules and index, L2 project knowledge, L3 topic-specific procedures; L1 is injected into model requests and details are read on demand	Long-lived projects with growing rules and knowledge

The agent maintains project memory according to the memory rules; routine manual organization of L1–L3 is not required. After saving, all existing and new projects use the selected mode from the next turn. Active turns and session execution workspaces are unchanged. Both memory stores are preserved, but are not automatically synchronized in full: Admin can read the official memory source and distill relevant knowledge as needed.

See User Quick Start: project modes (Chinese) for setup steps, memory files, and switching boundaries.

Choosing an Autonomy Mode

Three features keep an agent working without new input. They are not interchangeable:

Mode	Runs in	Decides "done"
Loop (chat rail)	The open chat session	A separate controller model, once per round
Goal Mode	A detached GA process with its own state dir	GA's `reflect/goal_mode.py`
UltraPlan (`/ultraplan`)	A single turn's tool loop	The main agent itself

Pick Loop when you want to watch and interrupt, and everything to stay in one thread; Goal Mode when the job is long and nobody needs to sit with it; UltraPlan when one multi-phase plan can be carried by the agent alone.

Loop spends one extra full-context controller call per round, so keep the round limit tight. It stops itself at the round limit, when the controller asks for the same next step twice in a row, or when you press stop.

Conductor: delegated work and delivery review

Use Conductor when a parent task should delegate bounded work to child sessions and collect their results. Unlike Loop's round controller, Conductor coordinates parent/worker sessions; execution completion is not delivery verification.

Session model-selection strategy: Tell the parent, for example, "Remember for this session: use model A for lookup and simple summaries, model B for complex design; ask me if unavailable." conductor_model_strategy supports set/get/reset; set fully replaces a policy of up to 8192 characters. A successful receipt applies it in the current turn, and it is freshly injected into every subsequent parent request, independently of trimmed history and after reopening. Save only user-provided or user-approved policy changes.

This is a routing preference, not an automatic backend router: the parent checks conductor_models and explicitly selects llm_no / reasoning_effort in conductor_dispatch. Explicit per-task choices take priority. Reset restores subtask defaults/inheritance. The policy does not alter the parent's model, permissions, other sessions, or queued/running workers, and is not written to global memory.

Subagent additional instructions: conductor_dispatch accepts additional_prompt (up to 8192 Unicode characters) for a single dispatch. conductor_defaults(action="set", additional_prompt="...") saves a default for future dispatches in this parent session. Omission inherits that default; an explicit empty string disables it for one dispatch. set preserves omitted defaults, additional_prompt: null clears only this default, and reset clears all defaults. Each worker persists the resolved prompt as a dispatch snapshot, including when queued; changing defaults does not modify existing workers. Reuse replaces the previous dispatch's additional prompt rather than accumulating it. The prompt is appended to existing instructions, never replaces system rules or grants permissions, and is not injected on ordinary user turns after a worker dispatch is terminal.

Project context is server-owned. New workers inherit the parent session project binding and workspace; the existing request-time resolver applies the current application project mode without rewriting session provenance. conductor_dispatch exposes no project selector. Legacy project_id arguments (including malformed values) are ignored by both adapter and server and cannot override this binding. Reuse preserves the worker context, but rejects an effective project or workspace mismatch with the parent before mutation; omit session_id to create a new inherited worker instead. Switching the global project mode follows the existing per-request behavior for both parent and worker, not a dispatch-specific migration.

Successful execution starts as Pending review. Verified (parent review) requires a concrete review basis; evidence_ids is optional. Explain the review method and supporting artifact paths, URLs, command output, or verification steps in basis, and state unverified boundaries. When citing the optional evidence ledger, copy IDs from conductor_collect's evidence[]; an empty ledger does not block a basis-only review. result_receipt is a read acknowledgment, not an evidence ID. Needs work means the delivery needs correction, not that evidence bookkeeping is absent. Worker claims or tool execution alone do not establish correctness, and parent review is not an independent guarantee. Single-worker tasks use lightweight review by default; high-risk or multi-worker work calls for stronger checks.
The parent workspace shows input/output tokens for the parent, finalized child snapshots, and their recorded total. Missing snapshots are not zero usage. Running usage is incomplete; child totals include only saved terminal snapshots, not live consumption or a billing estimate.
Each parent session accepts at most 48 cumulative dispatches, including reused workers and completed, failed, or cancelled dispatches. Dispatch 49 is rejected before session mutation. This is a dispatch-count limit, not a token, cost, or wall-clock budget. Existing concurrency (3 running) and nonterminal (12) limits still apply.

For Administrators

Goal Mode: Persistent goals (JSON), BBS team board, sync UI
File Operations: Browse GA root, edit skills/SOPs/configs, create/rename/delete
Models: Add/test/remove model profiles, wizard UI
Updates: Check GitHub Releases, download & apply platform-specific packages

For Developers

Frontend: React 18 + Vite 6, code-split routes, theme toggle, accessibility
Backend: Go 1.22+, embedded web assets (//go:embed web/dist), subprocess lifecycle
Build: Single-executable distribution, GitHub Actions CI/CD for 6 platforms
Test: npm run verify (lint + test:lib + build), go test ./...

🛠️ Development

CLI Flags

--headless / --server-only / --no-browser: Run without opening browser
--no-window: Use the system browser instead of the native desktop window (Windows and macOS; Linux always uses the browser)
--app-root <path>: Override GA root directory (default: from config.local.json)
--port <port>: Pin the listen port for this launch instead of the random loopback port

Environment Variables

GA_ADMIN_AUTH_USER / GA_ADMIN_AUTH_PASSWORD: Fix the credential for remote access. When both are set the password cannot be changed from the UI.
GA_ADMIN_NO_WINDOW: Same as --no-window

Remote access

The admin server can run processes and read and write files, so it stays on loopback unless you opt out:

Default. Binds 127.0.0.1 on a random port. Requests from this machine never need a password, and no other device can connect.
Remote access on, password required. Binds every interface on port. Remote clients must authenticate with HTTP Basic Auth; local requests still skip it. Set the password in Settings → Remote access first — a launch that requires a password without having one falls back to loopback and logs why.
Remote access on, anonymous allowed. Binds every interface with no authentication at all. Only appropriate on a network you fully trust.

Changes to the listen address take effect on the next start. Each run records where it actually bound in runtime.local.json (URL, address, port, PID); the file is removed on a clean shutdown and overwritten on the next start.

Configuration

Place config.local.json in the executable directory:

{
  "ga_root": "/path/to/GenericAgent",
  "remote_access": false,
  "remote_allow_anonymous": false,
  "port": 8787,
  "service_autostart": ["worker"],
  "slash_commands": [
    {"cmd": "/plan", "desc": "Call plan_worker.py for multi-step planning"}
  ]
}

host and port only apply while remote_access is true; a loopback launch always takes a random port. See config.example.json for all available options.

The repository ignores:

config.local.json
*.local.json
model_profiles.json
dist/
*.exe
web/node_modules/
/temp/
/release/
*.pid
*.log

Validation Before Commit

Run at least:

npm --prefix web run verify    # lint + test:lib + build
go test ./...
go build ./...
git diff --check

Notes:

npm run verify runs lint + test:lib + build (skips test:ui)
web/src/lib/*.test.mjs are auto-discovered by npm run test:lib
Test files do not need package.json registration
After changing internal/appicon/assets/tray_windows.ico, run go generate . to rebuild the committed rsrc_windows_*.syso files that give the Windows executable its icon

📦 Release

Verify clean state: No uncommitted changes, all tests pass
Run validation: npm --prefix web run verify && go test ./...
Commit & tag: git commit -am "release: v0.x.x" → git tag v0.x.x
Push: git push origin main --tags

GitHub Actions will build 6 platform packages and attach to the release:

ga-admin-windows-amd64.zip
ga-admin-windows-arm64.zip
ga-admin-linux-amd64.tar.gz
ga-admin-linux-arm64.tar.gz
ga-admin-darwin-amd64.tar.gz
ga-admin-darwin-arm64.tar.gz

Each package includes:

Platform-specific executable (ga-admin / ga-admin.exe)
config.example.json template
Version metadata (injected via -ldflags during build)

📄 Documentation

User-focused quickstart: docs/USER_QUICKSTART.md (Chinese)
Knowledge base: docs/knowledge_base.md
Developer experience: docs/secondary_dev_experience.md

🔗 Upstream

This project manages local GenericAgent instances. GA Admin requires a GenericAgent installation to function.

📄 License

This project is used internally within the GenericAgent ecosystem. For external distribution, confirm upstream GenericAgent and dependency project license requirements first.

🌟 项目简介

GenericAgent Admin Go 是 GenericAgent 的桌面管理面板。Go 后端负责进程、文件、配置、更新和系统集成，React/Vite 前端提供控制台界面。打包为单个 ga-admin / ga-admin.exe 可执行文件。

目标不是替代 GenericAgent，而是将本地 GA 运行状态、任务入口、模型配置、团队协作和桌面辅助整合到一个可维护的 UI 中。

🖥️ 界面预览

支持实例切换、模型选择、流式回复与用量跟踪的原生聊天界面。

📑 目录

核心特性
快速开始
核心功能
开发
发布
文档
上游项目
许可

📋 核心特性

特性	说明
📊 仪表盘	服务状态概览、最近活动、系统统计
💬 原生聊天	`/chat` 入口的聊天界面，流式响应，用量跟踪，模型切换
📝 文件编辑器	浏览 GA 根目录，编辑技能/SOP/配置，语法高亮，文件树
📋 任务管理	服务（启动/停止 worker）、计划任务、Goal 运行、自主报告
🧠 记忆浏览器	查看/搜索全局和项目记忆
📡 通道监控	活动通道、消息日志
🤖 自主模式	后台任务执行
📈 用量跟踪	每个模型的 Token/成本统计
🎯 Goal 模式	启动/停止长时间运行的目标，查看日志/状态
⚙️ 模型配置	添加模型的向导，测试端点，管理配置文件
🔧 设置	应用配置、认证、ga_root 路径
📄 日志查看器	尾随 worker 日志，搜索历史

📈 Star 历史

🚀 快速开始

⚠️ 前置要求： Python 3.11+（用于 GenericAgent），Node.js 18+ / Go 1.22+（用于开发构建）

给 LLM Agent 看的

获取安装指南并照做：

curl -fsSL https://raw.githubusercontent.com/Fwind43/GenericAgent-Admin/main/README.md

给人类用户看的

方法一 — 下载发布包（推荐）

从 GitHub Releases 下载平台特定包：

ga-admin-windows-amd64.zip
ga-admin-linux-amd64.tar.gz
ga-admin-darwin-amd64.tar.gz  (macOS Intel)
ga-admin-darwin-arm64.tar.gz  (macOS Apple Silicon)

解压后在同目录创建 config.local.json：

{
  "ga_root": "E:/Work/GenericAgent"
}

Windows： 双击 ga-admin.exe。界面会在原生桌面窗口中打开（基于 WebView2 运行时，Windows 11 与较新的 Windows 10 已预装），关闭窗口后程序继续驻留系统托盘。加 --no-window 可改用默认浏览器打开，加 --no-browser 则启动时不打开任何界面。 macOS： 运行 ./ga-admin。界面会在原生桌面窗口中打开（基于系统自带的 WKWebView）。关闭窗口后程序继续留在菜单栏。加 --no-window 可改用默认浏览器打开，加 --no-browser 则启动时不打开任何界面。 Linux： 运行 ./ga-admin 或 ./ga-admin --no-browser，Linux 仍使用默认浏览器打开界面。

默认监听 127.0.0.1 的随机端口：不对外暴露，本机访问也不需要密码。实际地址会在启动日志中打印，同时写入 runtime.local.json；需要再开一个界面视图时用它。要从其它设备访问，请在设置中开启远程访问（见远程访问）。

方法二 — 本地开发构建

cd GenericAgent-Admin-Go
npm --prefix web install
npm --prefix web run build
go run .

浏览器打开启动日志中给出的地址；若想固定本机端口，可运行 go run . --port 8787。Vite 开发服务器（npm --prefix web run dev）会把 /api 代理到 runtime.local.json 中记录的端口，因此随机端口也能自动跟上。

🎯 核心功能

面向日常用户

服务管理： 启动/停止 worker，监控日志，检查进程状态
聊天界面： /chat 入口，流式响应，用量跟踪，模型切换

项目模式怎么选

在聊天设置 → 项目模式 → 全局运行模式中选择模式并点击保存。这里改变的是项目记忆机制，不是模型或能力档位。

模式	记忆组织方式	选择建议
官方项目模式	沿用 GenericAgent 官方项目机制，使用项目的 `project_memory.md`	希望保持官方使用习惯时选择
Admin 项目模式（L1–L3）	L1 常驻规则与索引、L2 项目知识、L3 专题流程；L1 随模型请求注入，详细知识按需读取	适合长期积累、规则和知识较多的项目

项目记忆由 Agent 根据项目内对话维护，无需日常手动整理 L1–L3。保存后，已有项目和新项目从下一轮起使用所选模式，正在执行的轮次不变；切换不改变会话的执行工作目录。两套记忆分别保留，但不会自动全量同步：Admin 可参考官方旧记忆，按需提炼有效知识。

设置步骤、分层文件和切换边界见用户快速开始：项目模式。

三种自动推进模式怎么选

有三个功能都能让 Agent 在没有新输入的情况下继续干活，它们并不等价：

模式	运行位置	谁判断"做完了"
Loop（聊天右栏）	当前打开的会话内	独立的控制模型，每轮判一次
Goal 模式	独立的 GA 进程，自带状态目录	GA 自己的 `reflect/goal_mode.py`
UltraPlan（`/ultraplan`）	单个 turn 的工具循环内	主 Agent 自己

想边看边随时介入、并且产物都留在同一条会话里，用 Loop；任务长、不需要盯着，用 Goal 模式；一个多阶段计划 Agent 自己就能扛下来，用 UltraPlan。

Loop 每轮会额外花一次全量上下文的控制模型调用，轮次上限别设太大。它会在达到轮次上限、控制模型连续两次给出同一个下一步、或你手动停止时自行结束。

Conductor：任务分派与交付核验

需要父任务拆分工作、交给子会话执行并收集结果时使用 Conductor。它负责父子任务协作，不等同于 Loop 的逐轮控制；执行结束不代表成果已核验。

会话级模型选择策略： 可以直接告诉指挥家：“记住本会话的选模策略：查询和简单整理用模型 A，复杂设计用模型 B；模型不可用时先问我。”指挥家可用 conductor_model_strategy 的 set/get/reset 保存、查看或清除策略；set 完整替换策略，最多 8192 字符。保存后通过工具回执在当前轮生效，并在该父会话后续每次请求中重新注入系统提示，独立于对话历史裁剪，重开会话仍保留。只保存用户提供或授权修订的策略，不得自行覆盖。

策略是选模偏好，不是后端自动路由器：指挥家通过 conductor_models 查询现有模型，再在 conductor_dispatch 中显式选择 llm_no / reasoning_effort。单任务明确选模优先；清除策略后使用子任务默认值或继承规则。策略不改变父会话模型、权限、其他会话或已排队/运行的 worker，也不自动写入全局记忆。

执行成功后默认待核验。**已核验（父任务审阅）**必须提供具体的 basis，evidence_ids 可选：依据应说明审阅方式、产物路径、URL、命令输出或复验步骤，并写明未验证范围。引用工具证据时，只能使用 conductor_collect 返回的 evidence[] ID；证据列表为空也可依据实际交付进行轻验收。result_receipt 是已读回执，不是证据 ID。需返工表示交付质量需要修正，不表示缺少证据记账。worker 自述或工具执行本身不保证正确性，父任务审阅也不是独立的正确性保证。单 worker 任务默认轻验收，高风险、多 worker 或用户明确要求时加强核验。
父任务面板分列父任务、已封存子任务及已记录合计的输入/输出 Token。缺失快照不等于零消耗；运行中用量不完整，子任务仅计入已保存的终态快照，不是实时账单。
每个父会话累计最多接受 48 次派发，复用 worker 及已完成、失败、取消的派发均计数。第 49 次在修改会话前拒绝。这是派发次数上限，不是 Token、费用或运行时长预算；原有最多 3 个运行中、12 个未终结任务限制仍有效。

面向管理员

Goal 模式： 持久化目标（JSON），BBS 团队看板，同步 UI
文件操作： 浏览 GA 根目录，编辑技能/SOP/配置，创建/重命名/删除
模型管理： 添加/测试/移除模型配置文件，向导 UI
更新： 检查 GitHub Releases，下载并应用平台特定包

面向开发者

前端： React 18 + Vite 6，路由代码分割，主题切换，无障碍
后端： Go 1.22+，嵌入 web 资源（//go:embed web/dist），子进程生命周期
构建： 单可执行文件分发，GitHub Actions CI/CD 支持 6 平台
测试： npm run verify（lint + test:lib + build），go test ./...

🛠️ 开发

CLI 参数

--headless / --server-only / --no-browser：无浏览器模式运行
--no-window：改用系统浏览器而非原生桌面窗口（Windows 与 macOS；Linux 始终用浏览器）
--app-root <路径>：覆盖 GA 根目录（默认从 config.local.json 读取）
--port <端口>：本次启动固定监听端口，替代默认的本机随机端口

环境变量

GA_ADMIN_AUTH_USER / GA_ADMIN_AUTH_PASSWORD：固定远程访问使用的凭据；两者都设置后界面上无法再修改密码
GA_ADMIN_NO_WINDOW：等同于 --no-window

远程访问

管理端可以执行进程、读写文件，因此默认只监听本机，需要显式放开：

默认。 监听 127.0.0.1 的随机端口。本机请求永远不需要密码，其它设备也连不上。
开启远程访问 + 需要密码。 在所有网卡上监听 port。远程客户端必须通过 HTTP Basic Auth 认证，本机请求仍然免密。请先在设置 → 远程访问中设置密码；若配置要求密码却没有设置，启动时会退回本机监听并在日志中说明原因。
开启远程访问 + 允许匿名。 在所有网卡上监听且完全不做认证，只适合完全可信的网络。

监听地址的修改在下次启动时生效。每次运行都会把实际绑定信息（URL、地址、端口、PID）写入 runtime.local.json，正常退出时删除，下次启动时覆盖。

配置

在可执行文件目录放置 config.local.json：

{
  "ga_root": "/path/to/GenericAgent",
  "remote_access": false,
  "remote_allow_anonymous": false,
  "port": 8787,
  "service_autostart": ["worker"],
  "slash_commands": [
    {"cmd": "/plan", "desc": "调用 plan_worker.py 进行多步规划"}
  ]
}

host 与 port 只在 remote_access 为 true 时生效；本机监听始终使用随机端口。参见 config.example.json 获取所有可用选项。

仓库忽略：

config.local.json
*.local.json
model_profiles.json
dist/
*.exe
web/node_modules/
/temp/
/release/
*.pid
*.log

提交前验证

至少运行：

npm --prefix web run verify    # lint + test:lib + build
go test ./...
go build ./...
git diff --check

注意：

npm run verify 运行 lint + test:lib + build（跳过 test:ui）
web/src/lib/*.test.mjs 由 npm run test:lib 自动发现
测试文件无需 package.json 注册
修改 internal/appicon/assets/tray_windows.ico 后运行 go generate .，重新生成随仓库提交的 rsrc_windows_*.syso（Windows 可执行文件的图标资源）

📦 发布

验证清洁状态： 无未提交更改，所有测试通过
运行验证： npm --prefix web run verify && go test ./...
提交并打标签： git commit -am "release: v0.x.x" → git tag v0.x.x
推送： git push origin main --tags

GitHub Actions 将构建 6 个平台包并附加到发布：

ga-admin-windows-amd64.zip
ga-admin-windows-arm64.zip
ga-admin-linux-amd64.tar.gz
ga-admin-linux-arm64.tar.gz
ga-admin-darwin-amd64.tar.gz
ga-admin-darwin-arm64.tar.gz

每个包包含：

平台特定可执行文件（ga-admin / ga-admin.exe）
config.example.json 模板
版本元数据（构建时通过 -ldflags 注入）

📄 文档

用户快速开始：docs/USER_QUICKSTART.md（中文）
知识库：docs/knowledge_base.md
二次开发体验：docs/secondary_dev_experience.md

🔗 上游项目

本项目管理本地 GenericAgent 实例。GA Admin 需要 GenericAgent 安装才能运行。

📄 许可

本项目在 GenericAgent 生态系统内部使用。如需外部分发，请先确认上游 GenericAgent 及依赖项目的许可要求。

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
cmd		cmd
docs		docs
internal		internal
scripts		scripts
web		web
.gitignore		.gitignore
CHANGELOG.md		CHANGELOG.md
README.md		README.md
build.bat		build.bat
config.example.json		config.example.json
go.mod		go.mod
go.sum		go.sum
main.go		main.go
main_server_test.go		main_server_test.go
portable_bootstrap_other.go		portable_bootstrap_other.go
portable_bootstrap_windows.go		portable_bootstrap_windows.go
portable_bootstrap_windows_test.go		portable_bootstrap_windows_test.go
rsrc_windows_amd64.syso		rsrc_windows_amd64.syso
rsrc_windows_arm64.syso		rsrc_windows_arm64.syso

Latest commit

History

Folders and files

Repository files navigation