znlgis

面向 AI 编程助手的开源 GIS、CAD、C#、AI、IoT、3D 技能集合。覆盖 65 个开源项目，提供一站式技能索引、标签搜索、按需加载指南，让 AI 助手在 GIS/CAD/C#/AI/3D 领域获得更准确的代码生成与问题排查能力。

DocuTranslate（xunbu/docutranslate）是基于大语言模型的轻量级本地文件翻译工具，支持 PDF/Word/Excel/Markdown/EPUB/字幕/JSON 等多格式翻译，内置 MinerU PDF 解析、术语表自动生成、Web UI、REST API、Client SDK 与 MCP 接入。

AI/LLM 技能分类索引，覆盖 LLM 应用开发平台、自学习智能体、个人 AI 助手 Gateway、终端编码代理、多模型编排插件、文件翻译工具、AI 编程方法论等领域共 8 个开源项目，按需加载即可获得 Dify、Hermes Agent、OpenClaw、OpenCode、DocuTranslate 等工具的精准 AI 辅助。

geometry-api-net 是一个 .NET Standard 2.0 的空间几何计算库，提供对二维和三维几何对象的创建、解析、序列化以及空间关系运算能力。

GIS（地理信息系统）技能分类索引，覆盖空间数据处理、地图服务、几何运算、Web 可视化等领域共 23 个开源工具，按需加载即可获得 GDAL、QGIS、GeoServer、PostGIS、JTS 等工具的精准 AI 辅助。

robotgo-flow（znlgis/robotgo-flow）是基于 go-vgo/robotgo 构建的 Windows 桌面 RPA 自动化框架，通过声明式 YAML 定义工作流，利用图像模板匹配定位屏幕 UI 元素，自动执行鼠标/键盘/浏览器操作，内置人类行为模拟、交互式录制器与 .NET 10 WPF 托盘应用。

Others（其它）技能分类索引，覆盖 Go 语言核心语法与工程实践、RobotGo 跨平台桌面自动化（RPA）库、robotgo-flow 基于 YAML 的 Windows RPA 框架、自托管邮件营销平台 BillionMail、Java 微服务后台脚手架 RuoYi-Cloud 与 SSL 证书自动化客户端 acme.sh，按需加载即可获得邮件系统搭建、Spring Cloud 微服务开发、SSL/TLS 证书自动签发续期、Go 语言后端开发、GUI 自动化等通用工具的精准 AI 辅助。

3D 技能分类索引，覆盖 3D 高斯泼溅（3DGS）编辑与 Web 三维可视化领域，当前收录浏览器端开源高斯泼溅编辑器 SuperSplat，按需加载即可获得 .ply 高斯模型清理、裁剪、压缩、动画与发布的精准 AI 辅助。

Manage agent skills with gh skill. Use this skill to discover, preview, install, update, and publish Agent Skills so an agent can self-manage the skills available in its environment.

Patterns for invoking the GitHub CLI (gh) from agents. Use when the task mentions GitHub, gh, pull requests/PRs, issues, releases, gists, Actions/workflow runs, forks, repo cloning, reviews, or you need exact gh commands. Covers structured output, pagination, repo targeting, search vs list, issue types/sub-issues/discussions, projects, rulesets, and gh api fallback.

Token-frugal, multi-dimension code review for a diff/branch/PR. Use when reviewing changes, checking a PR, running a review→fix loop, or the task mentions "code review", "review my changes", "review this PR", "找 bug", "审查代码". Scales review depth to diff size, reports findings by dimension and severity, and calibrates against the project's threat model to avoid severity inflation. Reports findings; never rewrites code unless explicitly asked.

Run a lightweight, spec-driven change workflow — capture WHY/WHAT in a proposal, define WHAT the system does in specs, decide HOW in a design, break work into a tasks checklist, implement, then archive. Use when the task is a non-trivial feature or behavior change that benefits from durable, git-tracked planning artifacts, or when the request mentions "spec", "proposal", "propose a change", "spec-driven", "openspec", "design doc", "requirements", or "tasks checklist". Not for one-line fixes or pure Q&A.

Structured workflow for complex/heavy tasks that benefit from a review-gated, phased approach. Use when the task involves 3+ files, a major refactor, or architectural changes — or when the task mentions "deep work", "complex refactor", "multi-phase", "staged rollout". Creates durable artifacts (plan, verify, report) so the orchestrator can track progress across sessions.

Review recent work to identify recurring friction and propose minimal, durable improvements to the opencode config (agents, skills, commands, rules). Use when the user says "reflect", "what can we improve", "review our setup", "optimize the config", or after a series of related sessions reveals a pattern worth codifying.

Author and modify OpenCode configuration in this repository — opencode.json keys, agent/*.md frontmatter, skills/*/SKILL.md format, commands, and permissions. Use when editing opencode.json, adding or changing an agent, writing a skill or command, adjusting model routing/permissions, or the task mentions "opencode config", "agent prompt", "SKILL.md", "command", or "permission".

Generate a structured, hierarchical map of a repository or directory for quick orientation. Use when starting work in an unfamiliar codebase, when the task mentions "project structure", "codebase overview", "directory layout", "what does this repo contain", or when you need a bird's-eye view before diving into specific files. Produces a concise tree with annotations, saving repeated exploration tokens.