Building 120+ financial data MCP tools for AI agents: six design lessons from Equibles
DanielAPO · reddit · 2026-10-18
The author of Equibles, a remote MCP server for stock market data, shares design lessons for building MCP servers in data-heavy domains — the theme: make agents cite sources instead of guessing.
- Return source with every value: filing, period, date; three-call flow (SearchDocuments → SearchDocument → ReadDocumentLines) gets agents quoting the 10-K
- Spell out fiscal periods — "Q1 2027" differs for NVIDIA, Apple, Microsoft; responses state period end dates
- Tool descriptions capped at 500 chars act as the agent's index, pointing to the next tool
- MIC:TICKER codes for non-US stocks; reads and writes kept apart with action-named write tools
- OAuth first (dynamic client registration), API key fallback
Backend covers SEC filings, 13F holdings, insider trades, options, prices across US/Europe/HK/China. Free tier: 100 req/day; core is AGPL-3.0 open source.
More from coding & agent
- Jim Koppel poses a classic engineering problem: importing just one function from a bloated file — jimmykoppel · 2026-10-18
- The Long Walk: an interactive stroll through 300,000 years of human history, built with Runway MCP in Replit — tlakomy · 2026-10-18
- Stately.ai MCP earns early rave review from developer after live demo worked — TejasKumar_ · 2026-10-18
- Rust-based Git server on S3 handles 7,600+ concurrent pushes per second — blaizedsouza · 2026-10-18
- Chaining 2 PCs, a MacBook and 2 iPhones Runs 48-Expert MoE at 37 tok/s on 60k Context — StayLameBro · 2026-10-18
- Google, OpenAI, Meta among 13 applicants for .agent top-level domain — gaganghotra_ · 2026-10-18