LightWeightChartsMCP
Provides 6 MCP tools for semantic search, JS/TS code validation, and template generation based on TradingView Lightweight Charts documentation.
README
LightWeightChartsMCP
Complete TradingView Lightweight Charts reference documentation MCP server with 6 tools for semantic search, JS/TS code validation, and template generation.
Quick Start | Tools | Configuration | Development
pip install lwcmcp
lwcmcp
That's it. On first run, the server auto-builds the ChromaDB vector store from shipped JSON data (takes 30-60 seconds for embedding model download + indexing). Subsequent runs start instantly.
What you get:
- 6 MCP tools for Lightweight Charts docs lookup, JS/TS code validation, and template generation
- ~1617 entries indexed in a local ChromaDB vector store (100% offline)
- Sub-millisecond hot cache for priority lookups
- Regex-based JS/TS validator (no Node.js runtime required)
- Works with Claude Desktop, Cursor, Windsurf, OpenCode, and any MCP client
What you need:
- Python 3.10+
- Any MCP-compatible AI client
Features
- 6 tools: lookup, search, browse, validate, repair, scaffold
- 1 resource:
lwc://stats - 100% local: ChromaDB vector store, no network calls at runtime
- Fast: Hot cache for instant lookups, name index for fuzzy matching, semantic search (warm)
- 5 versions: 3.8, 4.0, 4.1, 4.2, 5.0 plus tutorials and framework integrations
- Validator: Regex-based JS/TS validation (no Node.js runtime required)
- 7 templates: basic_chart, realtime_updates, react_integration, vue_integration, web_components, custom_styling, multi_series
- Version filtering: Every tool accepts
versionparam to filter by LWC version - Production-ready: Health endpoint, circuit breakers, error masking, response capping, response caching middleware
Coverage
- LWC API: functions, interfaces, enumerations, type-aliases, variables (all 5 versions)
- Guides: getting-started, series-types, price-scale, time-scale, time-zones, plugins, migrations, android, ios, release-notes
- Tutorials: customization, a11y, react, vuejs, webcomponents, how-to, demos
- Total: ~1617 entries across 5 versions + tutorials
- ChromaDB: all entries indexed (JSON = ChromaDB, zero dedup loss)
Setup
pip install lwcmcp
Or from source:
git clone https://github.com/TheFractalyst/LightWeightChartsMCP.git
cd LightWeightChartsMCP
pip install -e ".[dev]"
Run the server
# stdio (for MCP clients like Claude, Cursor, opencode)
lwcmcp
# SSE (HTTP)
lwcmcp --transport sse --port 8080
# Build ChromaDB manually (auto-builds on first run otherwise)
lwcmcp build
MCP Client Configuration
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"lightweightchartsmcp": {
"command": "lwcmcp",
"transport": "stdio"
}
}
}
Cursor / Windsurf / OpenCode
Add to your project's .mcp.json or global MCP config:
{
"mcpServers": {
"lightweightchartsmcp": {
"command": "lwcmcp",
"transport": "stdio"
}
}
}
SSE (HTTP)
lwcmcp --transport sse --port 8080
Connect to http://localhost:8080.
Tools
+--------------+-----------+------------------------------------------+
| Tool | Type | Description |
+--------------+-----------+------------------------------------------+
| lwc_lookup | read-only | Exact name lookup with fuzzy fallback |
| lwc_search | read-only | Semantic search across all entries |
| lwc_browse | read-only | List namespace members |
| lwc_validate | read-write| Validate JS/TS for Lightweight Charts |
| lwc_repair | read-write| Fix common Lightweight Charts issues |
| lwc_scaffold | read-write| Generate chart template (7 kinds) |
+--------------+-----------+------------------------------------------+
+----------------+----------------------------------------------------+
| Tool | Signature |
+----------------+----------------------------------------------------+
| lwc_lookup | (name, kind?, version?) |
| lwc_search | (query, category?, namespace?, version?, n_results?)|
| lwc_browse | (namespace, category?, style?, version?) |
| lwc_validate | (code?, file_path?, file_content?, version?) |
| lwc_repair | (code, context, version?) |
| lwc_scaffold | (kind, name, description?, version?) |
+----------------+----------------------------------------------------+
Version Filtering
Every tool accepts an optional version parameter:
"3.8","4.0","4.1","4.2","5.0"- filter to specific versionNone(default) - search across all versions (prefers latest 5.0)
Configuration
+------------------------+------------------+--------------------------------------------+
| Env Var | Default | Description |
+------------------------+------------------+--------------------------------------------+
| TRANSPORT | stdio | Transport: stdio or sse |
| PORT | 8080 | Port for SSE transport |
| LWC_DB_PATH | ~/.lwc_mcp/db | ChromaDB path |
| LWC_COLLECTION | lwc_reference | ChromaDB collection name |
| LWC_EMBED_MODEL | all-MiniLM-L6-v2 | Sentence transformer model |
| LWC_MAX_RESULTS | 100 | Max results returned |
| VALIDATION_CACHE_TTL | 300 | Validation cache TTL (seconds) |
| VALIDATION_CACHE_SIZE | 500 | Validation cache max entries |
| LWC_INDEX_BATCH_SIZE | 25 | Indexing batch size |
| LAZY_MODEL | (empty) | Set to 1/true to defer model loading |
| LOG_DIR | ~/.lwc_mcp/logs | Log directory (SSE transport) |
| LOG_LEVEL | INFO | Logging level |
+------------------------+------------------+--------------------------------------------+
Namespaces
API Namespaces
+----------------+----------------------------------------------+
| Namespace | Description |
+----------------+----------------------------------------------+
| enumerations | Enums (LineStyle, CrosshairMode, etc.) |
| functions | Functions (createChart, addSeries, etc.) |
| interfaces | Interfaces (IChartApi, ISeriesApi, etc.) |
| type-aliases | Type aliases (SeriesType, Time, etc.) |
| variables | Variables and constants |
+----------------+----------------------------------------------+
Guide Namespaces
+------------------+------------------------------------------+
| Namespace | Description |
+------------------+------------------------------------------+
| getting-started | Getting started guides |
| android | Android platform integration |
| ios | iOS platform integration |
| price-scale | Price scale configuration |
| release-notes | Version release notes |
| series-types | Series types (candlestick, line, etc.) |
| time-scale | Time scale configuration |
| time-zones | Time zone handling |
| migrations | Version migration guides |
| plugins | Plugin system (v4.1+) |
+------------------+------------------------------------------+
Tutorial Namespaces
+----------------+----------------------------------------------+
| Namespace | Description |
+----------------+----------------------------------------------+
| customization | Chart customization tutorials |
| a11y | Accessibility tutorials |
| react | React framework integration |
| vuejs | Vue.js framework integration |
| webcomponents | Web Components integration |
| how-to | How-to guides |
| demos | Demo applications |
+----------------+----------------------------------------------+
Scaffold Templates (7 kinds)
+--------------------+----------------------------------------------+
| Kind | Description |
+--------------------+----------------------------------------------+
| basic_chart | Basic chart with createChart + setData |
| realtime_updates | Real-time data streaming with series.update |
| react_integration | React component with useRef + useEffect |
| vue_integration | Vue.js component with ref |
| web_components | Web Components custom element wrapper |
| custom_styling | Custom chart and series options/styling |
| multi_series | Multiple series (candlestick + volume) |
+--------------------+----------------------------------------------+
Development
ruff check . --line-length 120 # lint
pytest tests/ -v # test
Architecture
- FastMCP 3.4 with FileSystemProvider for auto-discovery
- ChromaDB vector store (cosine, all-MiniLM-L6-v2, 384-dim)
- 3 composable lifespans: db | model | cache
- Auto-build ChromaDB from shipped JSON on first run
- Response caching middleware (lookup/browse 1h, search 5m)
- Response limiting middleware (80KB cap)
- Circuit breaker on validator and ChromaDB
- Hot cache for fast lookups
- Error masking (no internal details leaked)
- Health endpoint at
/health
Database
- ~1617 entries across 5 versions (3.8, 4.0, 4.1, 4.2, 5.0) + tutorials
- Auto-built on first run from shipped JSON data (1.5MB)
- Stored at
~/.lwc_mcp/db(survives project updates)
To rebuild the database:
lwcmcp build
Data Pipeline
pipeline/scrape_lwc.py- Direct HTTP scraper for Lightweight Charts docs sitemap (github.io, no ScraperAPI needed)pipeline/merge_and_index.py- Merge + index into ChromaDB
Re-scrape from TradingView
pip install -e ".[pipeline]"
python pipeline/scrape_lwc.py
python pipeline/merge_and_index.py --reset
Tech Stack
- FastMCP 3.4 - MCP server framework with FileSystemProvider auto-discovery
- ChromaDB - Local vector database for semantic search
- SentenceTransformers -
all-MiniLM-L6-v2embedding model (384-dim) - RapidFuzz - Fuzzy string matching for name lookups
- Loguru - Structured logging with daily rotation (SSE mode)
Links
- fractalyst.dev - Author homepage
- TradingView - Public PineScript indicators
- @TheFractalyst - X / Twitter
License
MIT - see LICENSE
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.