# Give AI the tools, skills and knowledge to do useful work

Plan MCP servers, tool catalogs, reusable skills and governed knowledge so AI systems can work with current business context.

Canonical: https://aroai.com/solutions/mcp-skills-knowledge/

## Connect capability to the job

A tool gives an AI a bounded action or lookup. A skill provides the task process, terminology, constraints, examples and review requirements. A managed knowledge source supplies current, attributable business facts. Together they create a system the organisation can inspect and improve.

## What can be designed

- **MCP service portfolio**: Choose server boundaries, authenticated transports, tools, resources, prompts and client compatibility.
- **Skills and playbooks**: Capture job context, edge cases, definitions, examples and acceptance checks.
- **Knowledge ownership**: Set document source, authority, version, owner, permissions and review date.
- **Catalog and evaluation**: Maintain discoverable capability descriptions and tests that check the result quality.

## Useful starting point

Choose a workflow where experienced staff repeatedly explain how to work through a task. Capture its decisions and examples with the people who own it, then test the skill with realistic cases and visible sources.

## Product and delivery boundary

Aroai currently publishes MCP tools and an NZ company due diligence skill. Additional custom MCP services, sector skills and customer knowledge systems are scoped as delivery work or marked coming soon on the product roadmap; they are not included in the public tool catalog today.