For the complete documentation index, see llms.txt. This page is also available as Markdown.

Four Pillars

Short answer: Questflow’s financial harness has four pillars. Skills are trading strategies and investor judgment systems. Plugins are external data and tools. Models are the AI reasoning cores. Execution is the markets and platforms where the agent can read and trade. Together they turn fund manager judgment into an AI Finance Agent that can explain, monitor, and act inside your rules.

The map

Pillar
What it is
What it connects

Skills

Methodology — how the agent judges

Livermore, Buffett, Druckenmiller, VCP, crypto narrative, A-share systems, exclusive trade-labeled skills

Plugins

Data and tools — what the agent can see

Market data, on-chain, news, social, prediction markets, research

Models

Cognition — how the agent thinks

GPT, Claude, Gemini, open models, finance-specialized models

Execution

Venues — where the agent can act

Hyperliquid, Polymarket, OKX, Lighter, and expanding brokers

AI Finance Agent = Financial harness + Two-sided marketplace
Financial harness  = Models + Skills + Plugins + Execution Accounts

A model without skills is a chatbot. Skills without plugins are opinions. Plugins without execution stop at advice. Execution without judgment is a bot. The harness is the four together.

Why four, not one “AI”

Coding agents split the same way: the LLM, the skills/rules, MCP plugins, and the terminal. Finance needs the same split because data, venues, and judgment are fragmented. The harness is how Questflow closes the loop from “I understand the market” to a scoped order.

Related: What is a financial harness? · What is invest.md?