There are now two ways to wire quant research to data: classical HTTP REST API (your program calls it) and MCP — Model Context Protocol (let an LLM discover and call tools directly). They are complementary, not competitors. Here is how to choose.
Your code calls endpoints, parses JSON, handles auth:
r = requests.get(
"https://reachrich.ai/api/market/quote",
params={"code": "600519"},
headers={"Authorization": "Bearer ..."},
)
Model Context Protocol (Anthropic, late 2024) is an open standard that lets LLMs discover and invoke tools directly. For quant data:
rr_quote(code="600519") to get real-time priceStrengths: research-exploratory, conversation-driven, no programming barrier, composable across multiple MCP tools (data + execution + visualization). Best for: ad-hoc research, LLM agent workflows, prototyping.
| Layer | Mechanism | User |
|---|---|---|
| Programmatic | HTTP REST | Production strategies, automation |
| Conversational | MCP | Researchers, LLM agents |
ReachRich provides both: REST with token auth + MCP for direct Claude / Cursor / Cline access, sharing the same data contract.
Both paths need quotas, auth, and audit. Quant data is a commercial asset — bare endpoints invite scraping and abuse. Design with rate limits, per-key quotas, and per-tool MCP permissions from day one.