How It All Works
This portal gives your AI tools superpowers. Here's a plain-English guide to the three building blocks — and how they fit together.
The big picture
Skills
Teach the AI what to know and how to think
Subagents
Pre-configured AI personas with a specific role and guardrails
MCP Servers
Connect the AI to your live data and external tools
You can use any of these on their own, but they're most powerful together: a Subagent with the right Skills installed and an MCP Server connected to your data warehouse is a fully autonomous specialist.
“Think of a Skill like a recipe card slipped into your AI's cookbook.”
What it is
A Skill is a Markdown file that gives your AI assistant detailed, domain-specific instructions — how to think, what to watch for, and how to respond in a particular situation. On its own, your AI is a capable generalist. Add a Skill and it becomes a specialist.
How you use it
You install a Skill by dropping a SKILL.md file into your AI tool's skills folder (or using a one-line CLI command). The next time you open a chat, those instructions are automatically available. No code, no config — just drop the file and go.
Real example
For example, the "Clinical Coding Reviewer" skill teaches the AI ICD-10 and CPT coding guidelines, AHA Coding Clinic best practices, and exactly what to flag for compliance review. Without it, your AI gives generic answers. With it, it reviews like a certified coder.
“Think of a Subagent like a dedicated specialist on speed dial — a consultant who already knows your organization, your standards, and what they are responsible for.”
What it is
A Subagent is a pre-configured AI persona with a defined role, a specific set of tools it can (and cannot) use, and a custom system prompt that shapes how it thinks. It's not just a prompt — it's a fully packaged assistant ready to deploy.
How you use it
Subagents are defined in AGENT.md files. You load one into Auggie or Augment Code, and from that point on every interaction is shaped by the agent's role and constraints. You can even restrict which tools it is allowed to use — for example, a code review agent that can read files but never delete them.
Real example
The "Documentation Writer" agent knows healthcare writing standards, plain-language guidelines, and Medisolv documentation conventions. Hand it a care protocol draft and it improves clarity, checks reading level, and adds version metadata — no briefing required.
“Think of an MCP Server like a universal translator between your AI and the rest of your tech stack.”
What it is
MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external tools, databases, and APIs through a common interface. An MCP Server is a small service that sits between your AI and a data source, translating requests back and forth.
How you use it
You install an MCP Server once (usually a single npm or pip command). Your AI tool discovers it automatically and gains new capabilities — like querying your database, searching your wiki, or reading from an API — without you having to copy-paste data into the chat.
Real example
The "Snowflake MCP" server lets your AI run SQL queries against your clinical data warehouse and explain the results in plain English. Instead of exporting a CSV and pasting it into the chat, you just ask: "Which DRGs have the highest denial rate this quarter?"
Common questions
Which AI tools work with this portal?
Skills, Subagents, and MCP Servers are designed for Augment Code, Auggie CLI, and Snowflake Cortex Code. Each tool has slightly different install paths — the Skills Catalog shows exact instructions for all three.
Do I need to restart my AI tool after installing a Skill?
Usually no. Augment Code picks up new SKILL.md files on the next chat session automatically. Auggie and Cortex Code may require a reload of the tool window. Check the install tab on each Skill for tool-specific guidance.
Can I create my own Skills or Subagents?
Yes — and we encourage it. See the Contribution Guide in the footer for step-by-step instructions. All you need is a Markdown file and a pull request.
Is any of this sent outside Medisolv?
Skills and Subagent definitions are plain text files that live in your local AI tool. They shape how your AI thinks but do not transmit data anywhere on their own. MCP Servers connect to the data sources you configure — review each server's README for details on what data it accesses.
Ready to get started?