obai-hub

Routes financial-market questions to OBaI MCP servers and synthesizes grounded answers.

11|6|Updated Mar 18, 2026
One-click install
npx skills add https://github.com/sixteen-dev/obai --skill obai-hub-sixteen-dev
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: obai-hub
Source: https://github.com/sixteen-dev/obai/tree/main/skills/obai-hub
Command: npx skills add https://github.com/sixteen-dev/obai --skill obai-hub-sixteen-dev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Financial-market questions require live data that language models do not have in training data, and the OBaI platform spans ten separate MCP servers covering stocks, options, crypto, and prediction markets. This Skill acts as the central router: it maps each user question to the correct specialist server, enforces grounding rules so every numeric claim ties to tool output, and synthesizes multi-domain evidence into one coherent answer. ## Core Features & Use Cases - Intent Routing: A routing table maps questions about quotes, fundamentals, news, options, screening, portfolios, backtesting, research, prediction markets, and crypto to the correct MCP server on ports 8001-8010. - Grounding and Freshness Rules: Hard rules require MCP tool calls for live or numeric claims, prohibit speculation from training data, and define conflict-resolution and error-handling behavior. - Structured Synthesis: Output templates scale from short lookups to broad multi-domain analyses, while strategy, prediction-market, and crypto responses are delivered as terminal artifacts with their own contracts. - Use Case: A user asks to backtest a mean-reversion strategy on NVDA and compare it with current analyst sentiment; the hub resolves the ticker via the screening server, routes the backtest to the strategy server and the sentiment work to fundamentals, then merges the results into a single grounded report. ## Quick Start Ask a financial question such as "What is the current price and RSI of NVDA, and are there any upcoming earnings?" after registering the OBaI MCP servers from the bundled mcp-config.json.

Frequently Asked Questions about obai-hub

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I connect the OBaI MCP servers to my AI client?▼

Copy the bundled mcp-config.json into your project's .mcp.json, or register servers individually, for example with claude mcp add --transport http obai-market-data http://localhost:8002/mcp. Start the servers first with docker compose up -d from the repo root.

How does the hub decide which MCP server handles a financial question?▼

A routing table maps intent to server: quotes and technicals go to market-data on port 8002, fundamentals to 8001, options to 8004, backtests to 8007, prediction markets to 8009, and crypto to 8010. The matching specialist skill is read before any tool call.

Can the hub answer market questions without calling the MCP servers?▼

Only for static definitions or general finance concepts. Live, numeric, or time-sensitive claims must come from MCP tool calls because training data is stale; if a server is down, the hub reports the outage instead of answering from memory.

What happens when an OBaI server is down or a tool call fails?▼

The hub notes the unavailable data once, continues with verified data from other servers, and states how the gap limits the answer. It does not retry failed calls since the servers handle retries internally, and suggests docker compose up -d for down servers.

Where do user preferences like risk tolerance and initial capital live?▼

Preferences persist in ~/.obai/preferences.json with keys for risk tolerance, investment horizon, benchmark, initial capital, currency, and market. The hub reads this file when a task depends on a preference and updates it only when the user states a lasting preference.