Stock-signals MCP server: give LLMs conclusions, not raw data rows
op98765 · reddit · 2026-09-24
A developer shares vanoe-intelligence-mcp (MIT, on PyPI and the official MCP registry), an MCP server with 10 tools over a US stock/ETF market-signals API: composite verdict (0-100 score + rationale), trend stage with walk-forward weekly history, Point & Figure, FINRA short pressure, sector breadth, macro regime (FRED/CFTC), SEC insider + 8-K filings and more.
Design lessons worth stealing:
- Tools return conclusions, not rows: an LLM does arithmetic badly on 250 daily bars, but reasons well given "Stage 2, price 10% above a rising 30-week MA, score 85, here's why" — every factor carries a one-sentence quotable rationale.
- Every response declares staleness: asof, a freshness enum and suggestedrecheckseconds tell a looping agent when to re-ask instead of hammering the tool.
- Two transports, same tools: uvx stdio plus a hosted streamable-HTTP endpoint for URL-only clients (Claude.ai connectors, ChatGPT); keys ride in the path for header-less clients.
- Failures are free and typed: unknown ticker, upstream outage, rate limit all return structured errors with stable codes at no credit cost, so agents retry sanely.
Free key: 1,000 credits/month, no card required.
More from coding & agent
- Running models 24/7: we're in the harness phase, not autonomous builders yet — BLUECOW009 · 2026-09-24
- HeyGen guide: wiring Meta's Muse Agent to MCP for automated avatar videos — HeyGen · 2026-09-24
- Dev runs npm publishes and GitHub chores from a Tesla via Grok Bot — Baconbrix · 2026-09-24
- Grok Bot adds voice calls, 1Password and Slack drafts in a big feature week — mark_k · 2026-09-24
- Proof and Superfluid launch agent wallet verifying a human behind every transaction — csuwildcat · 2026-09-24
- Turning LLM hard classifiers into tunable soft classifiers with logprobs — JnBrymn · 2026-09-24