Connect Silicon Analysts to your AI stack
Six MCP tools — chip costs, wafer pricing, packaging, HBM, market pulse, and a chip-cost calculator — wired into the framework you already use. Each guide ships with working code you can paste into a script or a route handler today.
Anthropic SDK
Direct Claude integration via the MCP connector. Pass an mcp_servers block to the Messages API and Claude will discover all 6 tools automatically.
- ✓Native Claude tool use
- ✓Python and TypeScript examples
- ✓Serverless-compatible
LangChain
Use Silicon Analysts MCP tools inside any LangChain agent or LangGraph workflow via langchain-mcp-adapters. Mix with your own retrieval and search tools.
- ✓Works with create_react_agent
- ✓LangGraph compatible
- ✓Combines with custom tools
Vercel AI SDK
Stream Silicon Analysts MCP tools into Next.js, SvelteKit, or any TypeScript app via experimental_createMCPClient. Provider-agnostic — Anthropic, OpenAI, Google.
- ✓streamText + generateText
- ✓Next.js App Router native
- ✓Provider-agnostic models
What every integration gives you
Six MCP tools, one connection
get_accelerator_costs— 13 AI chips, COGS & margincalculate_chip_cost— die dimensions → costget_hbm_market_data— 9 HBM tablesget_market_pulse— supply-chain signalsget_wafer_pricing— prices by nodeget_packaging_costs— CoWoS, EMIB, SoIC, FC-BGA
Provenance on every record
Every tool response carries a provenance block:
last_updated— ISO 8601source_type— research / derived / computed / estimatedconfidence_tier— high / medium / lowdataset_version
See /data-quality for the canonical schema.
Don’t see your stack?
The Silicon Analysts MCP server is a standard Model Context Protocol streamable HTTP endpoint and works with any MCP client. The base URL is https://siliconanalysts.com/api/mcp with an Authorization: Bearer header.
Prefer plain REST? The same data is available at /api/v1/* — see the Developer API docs for endpoints, schemas, and curl examples.