Skip to main content
Alchemist exposes a full Model Context Protocol (MCP) server. Any MCP-compatible agent can call Alchemist’s search and document tools without writing any integration code — you just point the agent at the endpoint.

​
MCP endpoint

https://api.askalchemist.com/mcp/
Authentication: Authorization: Bearer alch_YOUR_KEY
Transport: Streamable HTTP (MCP 2025 spec)

​
Available tools

ToolWhat it does
search_insightsPrimary search — semantic query over the full document database
get_documentFetch a document and all its insights by doc_id
list_documentsBrowse documents by source or date
list_filter_optionsGet all valid source_name values
search_webFallback live web search when the database has no results

​
Shared configs — configure once

Several agent surfaces share configuration, so you may not need to set things up more than once:
If you configure hereIt also works in
claude.ai Connectors UIClaude Code CLI (auto-syncs on same Anthropic account)
Claude Desktop claude_desktop_config.jsonClaude Code CLI (import with claude mcp add-from-claude-desktop)
.mcp.json in your project rootClaude Code CLI, Claude Code web, VS Code extension
Cursor .cursor/mcp.jsonRoo Code, Continue, and other VS Code MCP agents (same format)

​
Choose your integration

Claude

Claude.ai web, Claude Desktop, and Claude Code — configure once, sync everywhere.

Cursor

Cursor IDE — one JSON block in .cursor/mcp.json.

ChatGPT

ChatGPT Custom GPT Actions — no MCP required, REST-based.

OpenClaw

OpenClaw open-source personal AI assistant — MCP via openclaw mcp add.

Codex CLI

OpenAI Codex CLI — TOML config at ~/.codex/config.toml.

Hermes

Nous Hermes and other local models via Ollama or LM Studio.

Generic MCP

Any MCP-compatible client — copy-paste config and system prompt.

​
Manual integration (non-MCP)

If your agent doesn’t support MCP, integrate directly via the REST API: 
You have access to Alchemist — a financial research database of primary source documents
(FOMC minutes, USDA WASDE, SEC filings, economic releases, etc.).

To search it, call:
  POST https://api.askalchemist.com/api/insights/search
  Header: X-API-Key: alch_YOUR_KEY
  Body: {"query": "<focused 3-8 word phrase or question>", "source_name": null, "since": null, "limit": 20}

To get a full document:
  GET https://api.askalchemist.com/api/documents/{doc_id}
  Header: X-API-Key: alch_YOUR_KEY

To list available sources:
  GET https://api.askalchemist.com/api/documents/filters
  Header: X-API-Key: alch_YOUR_KEY

Search guidance:
- Phrase queries as one focused concept (3–8 words or a concise question).
- Avoid keyword soup and conversational wrappers like "tell me about".
- Run 1–2 searches, read results, then refine.
- Retry with different phrasing before concluding the database has nothing.
- For deep research, find a doc_id via search, then fetch that document for full context.
- Call /api/documents/filters first when you need to filter by source.