Common AI Knowledge
This page collects short reference notes about common AI concepts used in this site.
Prompt
Prompt engineering notes and patterns:
- Prompt = input + instructions + context.
- Keep context concise and structured.
- Use system and assistant roles when supported.
Examples and templates:
- Template 1: "You are an expert X. Given the following input: {{input}}. Respond with..."
- Template 2: Short few-shot examples.
Agent
A short definition:
- Agent: an autonomous system that perceives its environment, makes decisions, and acts to achieve goals. In AI, agents may be simple rule-based programs, reinforcement learning agents, or orchestration layers that call models and tools.
MCP
- MCP: Model Context Protocol (or "Model Control Plane" depending on context). Use this section to list what MCP means for you — orchestration, connectors, observability, and dataflow.
Example:
- Purpose: coordinate multiple models, maintain context, routing and plugin/tool control.
Combined overview of prompts, MCP and agents.
