lampa-mcp-server
MCP server for AI-assisted development on the Lampa open-source TV app, providing read-only access to the source tree with 41 tools and 5 resources for structured analysis, planning, editing, and validation.
README
lampa-mcp-server
An MCP server for AI-assisted development on the Lampa open-source TV app.
It gives AI agents (Claude, Cursor, etc.) structured, read-only access to the Lampa source tree — so they understand the repo before making changes.
What it does
The server exposes 57 tools and 6 resources across nine capability layers:
| Layer | File | Tools | Purpose |
|---|---|---|---|
| Discovery | discovery.ts |
6 | Navigate repo structure, list modules, search code, read files |
| Analysis | analysis.ts |
7 | Locate settings, API calls, UI components, translations, dependency maps |
| Planning | planning.ts |
4 | Generate change plans, impact analysis, edit targets, risk scans |
| Editing | editing.ts |
4 | Draft patches, scaffold plugins, generate hook/setting boilerplate |
| Validation | validation.ts |
4 | Quality checks, find tests, resolve build commands, query docs |
| Lampa Deep | lampa_deep.ts |
8 | Deep Lampa-specific analysis: providers, events, translations, lifecycle |
| Advanced | advanced.ts |
8 | File reads, storage schema, event bus map, network map, pattern guide |
| Lampa Modern | lampa_modern.ts |
10 | Maker modules, CUB API (Lampa subset), WebSocket, components, settings, packaging |
| CUB API | cub.ts |
6 | Full CUB API surface from Lampa source: catalog, auth, sync, data models |
Tools
Discovery
repo_overview— summarise app structure, folders, entrypoints, plugins, and scriptslist_modules— list JS/TS modules in any subfolderfind_files— find files by name pattern or extensionsearch_code— regex/text search with file:line previews (usesripgrepwhen available)read_file_segment— read a specific line range from any repo filelist_scripts— show all NPM scripts frompackage.json
Analysis
find_settings— locateLampa.Settings.addandLampa.Storageusage; optionally filter by keywordfind_api_calls— findfetch,$.ajax, and provider integrationsfind_ui_component— find templates, components, and views by namefind_translation_keys— look up translation key definitions and usages across all language filesfind_styles_for_module— find CSS/SCSS files related to a module or feature namemodule_dependency_map— map imports and reverse dependencies with change blast radiusfind_feature— infer all files relevant to a named feature (player, catalog, iptv, search, etc.)
Planning
plan_feature_change— step-by-step implementation plan (call beforedraft_patch)impact_analysis— reference count and risk level for a target filesuggest_edit_targets— minimal file set and safe insertion guidance for a feature requestrisk_scan— detect coupling risks: global events, storage, shared templates, DOM globals
Editing
draft_patch— guided patch draft with file previews and Lampa patterns (requires priorplan_feature_change)insert_hook— find the bestLampa.Listenerhook point for a lifecycle triggeradd_setting— generate toggle / select / input setting boilerplate with storage wiringscaffold_plugin_integration— full plugin folder scaffold withmain.js, CSS, and component boilerplate
Validation
run_grep_checks— scan for TODOs,console.logleftovers, looseundefinedchecks, hardcoded stringslist_related_tests— find spec files related to a module or featurerun_build_hint— resolve the right build / dev / test / lint command frompackage.jsondoc_lookup— search generated docs or README for a topic
Lampa Deep
plugin_deep_dive— single-call analysis of a plugin folder: files, Lampa API usage, events, settings, entry-point previewlist_streaming_providers— catalog all online streaming providers with base URLs, methods, and Lampa APIs usedtranslation_coverage— compare all language files against the English reference; shows coverage % and missing keys per languagetrace_event— trace aLampa.Listenerevent through the full codebase: who sends it, who follows itlampa_api_surface— extract the completeLampa.*global API surface: every module with sub-methods and file usage countslist_templates— list all UI templates insrc/templates/; optionally read one to inspect HTML and data bindingsgenerate_plugin_boilerplate— generate working plugin code for selected features: settings, hooks, storage, lang keys, IPTVcomponent_lifecycle— deep-analyse a component: lifecycle methods, events, APIs, storage, templates, and settings
Advanced
read_file— read a complete file (truncated atmax_lines); useread_file_segmentfor large filesget_storage_schema— scan allLampa.Storage.get/setcalls; returns a complete key → default / readers / writers tablelist_all_events— build a fullLampa.Listenerevent bus map: every event, listener count, emitter count, orphan detectionget_network_map— extract all hardcoded URLs, base URL variables, and proxy names from any scopevalidate_plugin— score a plugin against 10 Lampa conventions (IIFE, strict mode,appreadybootstrap, noeval, etc.)extract_template_html— extract actual HTML markup, CSS classes, and data-binding placeholders fromsrc/templates/*.jsget_core_module— browse and readsrc/core/modules; lists all available when called without a nameexplain_lampa_pattern— pattern reference guide backed by live source examples:iife-plugin,storage,settings,events,component,request,template,activity,player-hook
Lampa Modern (v3.0 architecture)
maker_module_map— map Maker classes (Card, Main, Category, Line, …), module/map files, and lifecycle hookscub_api_catalog— catalog all CUB cloud REST endpoints with method and source filesocket_protocol_map— WebSocket inbound/outbound methods, mirrors, and payload envelopeactivity_component_registry— allComponent.addregistrations and Router routeslampa_settings_flags—window.lampa_settingsfeature flags and platform overridesplatform_packaging_guide— gulp targets for web, webOS, Tizen, GitHub Pages, plugins, docscontent_rows_api— allContentRows.addhome-screen injection pointsfavorite_category_schema— favorite/bookmark category types and timeline marksmanifest_mirrors_map— cub_mirrors, soc_mirrors, cub_domain resolution logicupgrade_migration_checker— detect deprecated 2.x APIs (Lampa.Card, InteractionMain, etc.) in a file
CUB API (from Lampa source — CUB developer docs)
cub_api_catalog— full CUB API catalog extracted from Lampa (account, bookmarks, timeline, AI, collections, TMDB proxy, plugins, WebSocket)cub_endpoint_detail— single endpoint with source file, code context, feature gatescub_auth_guide— device/add login, token/profile headers, Permit gating, Premium, mirrorscub_data_models— bookmark types, timeline storage, favorite categories, sync shapescub_sync_guide— bookmark/timeline/storage sync: REST dump/changelog + WebSocketcub_timeline_hash_guide— Utils.hash() algorithm for watch progress keys
Resources
| URI | Description |
|---|---|
repo://overview |
Top-level directory listing |
repo://scripts |
NPM scripts as JSON |
docs://index |
Generated JSDoc (requires npm run doc in the Lampa repo) |
settings://catalog |
All Lampa.Settings.add registrations in the repo |
api://integrations |
All API and network call sites |
cub://lampa-api |
CUB API endpoints JSON extracted from Lampa source |
Requirements
- Node.js 20+
- A local checkout of the Lampa source repo
- Optional:
ripgrep(rg) for faster search — falls back to pure Node if not present
Setup
git clone https://github.com/your-username/lampa-mcp-server
cd lampa-mcp-server
npm install
npm run build
Set the path to your Lampa checkout via the environment variable:
export LAMPA_REPO_PATH=/path/to/lampa-source
Or pass it inline when starting the server:
LAMPA_REPO_PATH=/path/to/lampa-source node dist/index.js
Connect to Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"lampa-mcp-server": {
"command": "node",
"args": ["/absolute/path/to/lampa-mcp-server/dist/index.js"],
"env": {
"LAMPA_REPO_PATH": "/absolute/path/to/lampa-source"
}
}
}
}
Connect to Cursor
Add to .cursor/mcp.json in your project (or the global ~/.cursor/mcp.json):
{
"mcpServers": {
"lampa-mcp-server": {
"command": "node",
"args": ["/absolute/path/to/lampa-mcp-server/dist/index.js"],
"env": {
"LAMPA_REPO_PATH": "/absolute/path/to/lampa-source"
}
}
}
}
Recommended agent workflow
The server enforces a deliberate two-step contract: plan first, patch second.
repo_overview → find_feature → module_dependency_map
→ plan_feature_change → draft_patch
For deep plugin work, start with the single-call analysis tools:
plugin_deep_dive → maker_module_map → cub_api_catalog
→ plan_feature_change → draft_patch → validate_plugin
For CUB account/sync work:
cub_auth_guide → cub_api_catalog → cub_sync_guide
→ cub_data_models → cub_endpoint_detail
System prompt for best results:
You are an AI coding agent working on the Lampa source repository. Always begin by collecting context through MCP tools: overview, relevant modules, settings, API calls, and dependency map. Do not invent project structure or framework patterns. Before writing code, produce: affected files, why each matters, implementation steps, risks, and validation checks. When generating code, preserve naming, style, and surrounding patterns used in the target files.
Project structure
src/
├── index.ts # Server entry point
├── config.ts # LAMPA_REPO_PATH + CUBCTL_PATH config
├── utils/
│ ├── fs.ts # File system helpers
│ ├── search.ts # ripgrep + Node fallback search
│ ├── lampa.ts # Lampa-specific patterns, feature map, risk patterns
│ ├── lampa_deep.ts # Deep analysis utilities (events, lifecycle, providers)
│ ├── lampa_modern.ts # Lampa 3.0 architecture extractors (Maker, CUB subset, socket)
│ └── cub.ts # CUB API extraction from Lampa source (scanner + metadata)
├── tools/
│ ├── discovery.ts # Phase 1 — repo navigation (6 tools)
│ ├── analysis.ts # Phase 2 — Lampa understanding (7 tools)
│ ├── planning.ts # Phase 3 — change planning (4 tools)
│ ├── editing.ts # Phase 4 — assisted editing (4 tools)
│ ├── validation.ts # Phase 5 — quality checks (4 tools)
│ ├── lampa_deep.ts # Lampa-specific deep analysis (8 tools)
│ ├── advanced.ts # File I/O, schema, pattern guide (8 tools)
│ ├── lampa_modern.ts # Lampa 3.0 architecture tools (10 tools)
│ └── cub.ts # CUB API extraction from Lampa source (6 tools)
└── resources/
└── index.ts # MCP resources (6 read-only stable context endpoints)
Development
npm run build # compile TypeScript → dist/
npm run dev # run directly with ts-node
npm start # run compiled dist/index.js
npm run typecheck # type-check without emitting
npm run lint # ESLint
npm run lint:fix # ESLint with auto-fix
npm run format # Prettier write
npm run format:check # Prettier check (CI)
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。