lunw/shopline-ai-toolkit-dsh
SHOPLINE AI Toolkit for DeepSeek Harness (dsh-plugin): official SHOPLINE Developer MCP bridge + SHOPLINE agent skills, mirroring the Shopify AI Toolkit architecture. dsh-plugin
Listed
3
Skill
Bundle verified
Preview
What it does
SHOPLINE AI Toolkit: bridges the official SHOPLINE Developer MCP server and ships seven SHOPLINE agent skills (Admin REST, GraphQL, OAuth, webhooks, Sline) — the SHOPLINE counterpart of the Shopify AI Toolkit.
Best for
- SHOPLINE developers using DeepSeek Harness who need platform-aware coding assistance.
- Workflows involving SHOPLINE Admin REST, GraphQL, OAuth, webhooks, or Sline templates.
- Agents that need to search official SHOPLINE documentation or inspect and validate API definitions through MCP.
Not ideal for
- Shopify or other commerce-platform development that does not target SHOPLINE.
- Users outside DeepSeek Harness who only need a general-purpose API client.
- Workflows requiring the optional community Sline MCP without additional configuration, because it is disabled by default.
README
SHOPLINE AI Toolkit for DeepSeek Harness (DSH)
🛍️ SHOPLINE AI Toolkit — a DeepSeek Harness (dsh) plugin that brings SHOPLINE 开发者 MCP 官方工具与 SHOPLINE 平台知识技能 到你的 AI 编程助手。 架构参照 Shopify AI Toolkit:
Skills + Dev MCP Server + Agent Plugin。dsh-plugin · SHOPLINE · MCP · AI Toolkit · DeepSeek Harness
English
What is this?
The Shopify AI Toolkit gives AI coding agents Shopify-aware context through three layers: agent skills, a Dev MCP server, and per-tool plugins. This project brings the same architecture to SHOPLINE developers working inside DeepSeek Harness (dsh):
| Layer | What you get |
|---|---|
🧠 Agent skills (skills/) |
7 SKILL.md bundles: shopline-onboarding-dev, shopline-admin-rest, shopline-graphql, shopline-oauth, shopline-webhook, shopline-sline, shopline-dev-mcp — endpoint URLs, signing rules, error codes, retry policies, all grounded in developer.shopline.com docs and community experience
|
| 🔌 Dev MCP bridge | The official SHOPLINE Developer MCP (@shoplineos/shopline-developer-mcp, local stdio, no auth) is wired through dsh’s native MCP client — tools appear as mcp__shopline__search_shopline_docs, mcp__shopline__get_admin_rest_endpoint_detail, mcp__shopline__get_graphql_schema, mcp__shopline__validate_graphql_codes, … |
| 🎛️ Optional community MCP |
sline.dev server (@shopline/dev-mcp) adds mcp__sline__introspect_admin_schema, mcp__sline__validate_theme, … (commented out by default) |
| 📦 dsh bundle |
cordis.patch.yml + lib/index.js, installable with dsh plugin --profile <name> add shopline-ai-toolkit-dsh
|
Install
See docs/install.md. Quick start (full bundle):
dsh plugin --profile web add shopline-ai-toolkit-dsh
# add "shopline-ai-toolkit-dsh" to dsh.profile.bundles in ~/.dsh/profiles/web/package.json
dsh web
Skills-only: cp -R skills/* ~/.dsh/skills/
Example
“Search the SHOPLINE Admin REST endpoints for products, then give me the full endpoint detail for updating a product variant, including request parameters and a curl example.”
The agent loads shopline-admin-rest + shopline-dev-mcp skills and calls
mcp__shopline__search_admin_rest_endpoints → mcp__shopline__get_admin_rest_endpoint_detail.
More prompts: examples/prompts.md · Design: docs/architecture.md
Repo layout
├── package.json # dsh bundle manifest (dsh.bundle.patch)
├── cordis.patch.yml # bundle patch: mcp-shopline bridge + shopline-toolkit plugin
├── lib/index.js # Cordis plugin: registers skills/* on ctx.skills
├── skills/ # 7 SHOPLINE agent skills (SKILL.md bundles)
├── mcp/mcp.shopline.json # ready-to-use MCP client configs (Cursor/Claude/DSH)
├── docs/ # architecture.md · install.md
└── examples/prompts.md # copy-paste prompts (EN/中文)
License
MIT. Not affiliated with SHOPLINE; official docs and npm packages remain the source of truth.
简体中文
这是什么?
Shopify AI Toolkit 通过三层结构让 AI 编程助手获得 Shopify 平台上下文:agent 技能 (skills)、 Dev MCP 服务器、以及各 AI 工具的插件。本项目把同样的架构带给在 DeepSeek Harness (dsh) 中开发的 SHOPLINE 开发者:
| 层 | 内容 |
|---|---|
🧠 Agent 技能(skills/) |
7 个 SKILL.md 技能:shopline-onboarding-dev(入门)、shopline-admin-rest(REST API)、shopline-graphql(GraphQL)、shopline-oauth(授权与签名)、shopline-webhook、shopline-sline(模板引擎)、shopline-dev-mcp(MCP 工具用法)。端点、签名规则、错误码、重试策略均来自 developer.shopline.com 官方文档与 community.shoplineapp.cn 社区实践 |
| 🔌 Dev MCP 桥接 | 通过 dsh 原生 MCP 客户端接入官方 SHOPLINE 开发者 MCP(@shoplineos/shopline-developer-mcp,本地 stdio、免鉴权)。工具以 mcp__shopline__search_shopline_docs、mcp__shopline__get_admin_rest_endpoint_detail、mcp__shopline__get_graphql_schema、mcp__shopline__validate_graphql_codes 等形式直接可用 |
| 🎛️ 可选社区 MCP |
sline.dev 服务器(@shopline/dev-mcp)提供 mcp__sline__introspect_admin_schema(Admin GraphQL Schema 内省)、mcp__sline__validate_theme(Sline 模板校验)等,默认注释、按需开启 |
| 📦 dsh 插件包 |
cordis.patch.yml + lib/index.js,一条命令安装:dsh plugin --profile <name> add shopline-ai-toolkit-dsh
|
安装
详见 docs/install.md。快速开始(完整插件):
dsh plugin --profile web add shopline-ai-toolkit-dsh
# 在 ~/.dsh/profiles/web/package.json 的 dsh.profile.bundles 中加入 "shopline-ai-toolkit-dsh"
dsh web
只想用技能:cp -R skills/* ~/.dsh/skills/
示例
「搜索 SHOPLINE Admin REST 端点中商品相关接口,然后给出更新商品变体的完整端点详情, 包括请求参数和 curl 示例。」
助手会加载 shopline-admin-rest 与 shopline-dev-mcp 技能,并依次调用
mcp__shopline__search_admin_rest_endpoints → mcp__shopline__get_admin_rest_endpoint_detail。
更多示例: examples/prompts.md · 设计文档:docs/architecture.md
仓库结构
├── package.json # dsh bundle 清单(dsh.bundle.patch)
├── cordis.patch.yml # bundle patch:mcp-shopline 桥接 + shopline-toolkit 插件
├── lib/index.js # Cordis 插件:把 skills/* 注册到 ctx.skills
├── skills/ # 7 个 SHOPLINE 技能(SKILL.md 包)
├── mcp/mcp.shopline.json # 开箱即用的 MCP 客户端配置(Cursor/Claude/DSH)
├── docs/ # architecture.md · install.md
└── examples/prompts.md # 可直接复制的提示词(中/英)
许可证
MIT。本项目与 SHOPLINE 无隶属关系;官方文档与 npm 包始终是权威依据。
Frequently Asked QuestionsFAQ
Use the verified command dsh plugin --profile default add github:lunw/shopline-ai-toolkit-dsh in a DSH-enabled shell. The command resolves the public package metadata and keeps the plugin attached to the catalog identity shown on this page.
Compatibility follows the bundle and profile status shown above. If a profile is not detected, keep the plugin disabled there and check the repository documentation before enabling it in production.
The GitHub link and activity metadata are the source of truth for releases and maintenance. Revisit this page after a new release to confirm the catalog has observed the latest version.