AI Skill - Flinks Documentation

The Flinks AI Skill is a curated markdown file that gives AI coding assistants everything they need to build a Flinks integration in one shot: authentication flows, Connect setup, event handling, data retrieval with polling, MFA handling, error codes, and test credentials. Skill URL: https://docs.flinks.com/skill.md

What is an AI Skill?

An AI skill is a structured markdown document designed specifically for AI consumption. Unlike an MCP server that provides real-time search across documentation, a skill delivers the complete integration context upfront, allowing an AI assistant to build a working implementation from a single prompt. The Flinks skill covers:

Authentication flow: Token generation, authorize calls, credential handling
Flinks Connect setup: Iframe parameters, event listeners, redirect handling
Data retrieval: GetAccountsDetail, polling on 202, async endpoints
MFA handling: 203 responses, security challenges, re-authorization
Refresh flows: Cached vs live re-authorization, nightly refresh setup
Error codes: 401, 202, 203 responses and recovery actions
Enrichment: Income, lending, and credit risk attributes
Test users: Sandbox credentials and demo institution

How to Use

With Claude Code

Add the skill as a local file or fetch it directly:

# Download the skill
curl -o flinks-skill.md https://docs.flinks.com/skill.md

# Reference it in your prompt
claude "Using the Flinks skill in ./flinks-skill.md, build a bank account \
  linking flow with React frontend and Express backend"

With Any AI Assistant

Fetch the skill file from https://docs.flinks.com/skill.md
Include it in your AI assistant’s context (paste it, attach it, or reference it)
Describe what you want to build. The skill provides enough context for a complete implementation

Example Prompt

Using the Flinks skill context in ./flinks-skill.md, implement a full Flinks
bank account aggregation app. React + Tailwind for the frontend (Vite), FastAPI
for the backend. The app should: (1) generate an authorize token on the backend,
(2) load Flinks Connect iframe with the token, (3) capture the loginId via event
listener, (4) call /Authorize then /GetAccountsDetail with polling on 202,
(5) handle 203 MFA (Connect handles it in-iframe during initial connect),
(6) display account data with balances and transactions, (7) offer a "Refresh
Data" button that triggers a live re-authorization. If the backend gets 200,
fetch fresh GetAccountsDetail data; if 203, return the requestId + a fresh
authorize token to the frontend so it can relaunch Flinks Connect iframe with
?requestId=...&authorizeToken=... for the user to answer MFA, then fetch data
after redirect. Use sandbox credentials.

Skill vs MCP Server

	AI Skill	MCP Server
What it is	Single markdown file with complete context	Live search across all documentation
Best for	Building new integrations from scratch	Looking up specific details during development
How it works	One-shot context: AI reads it all upfront	On-demand queries: AI searches as needed
Setup	Download a file or paste a URL	Configure server connection in your IDE
Use when	”Build me a Flinks integration"	"What parameters does this endpoint accept?”

Use the skill when you want an AI to build something end-to-end. Use the MCP server when you need targeted answers during ongoing development. They work well together: start with the skill for the initial build, then use MCP for refinements.

Resources

Skill file: the raw markdown file
llms.txt: comprehensive page listing for AI navigation
MCP Server Setup: configure real-time documentation search

​What is an AI Skill?

​How to Use

​With Claude Code

​With Any AI Assistant

​Example Prompt

​Skill vs MCP Server

​Resources