A self-paced course on the frontend side of agentic systems: the interfaces through which people watch, steer, approve and correct AI agents. Six modules cover agent interaction patterns, streaming and token-level rendering, visualising tool calls and multi-step plans, approval and interruption flows, client-side state and memory, and evaluating and shipping agent interfaces responsibly.

Written for frontend and full-stack developers who already know a component framework and are now asked to put an agent in front of users. Tech Skills Library is independent and issues no certification, credential or licence.

Modules

Module One — Agent Interaction Patterns in the Browser

Lessons

• Chat Is Not the Only Agent Interface Preview
• Mapping the Agent Loop to User Intent
• Latency Budgets for Agentic Frontends

Module Two — Streaming and Token-Level Rendering

Lessons

• Server-Sent Events Versus WebSockets for Agent Output
• Rendering Partial Markdown and Code Safely
• Backpressure and Cancellation in the Client

Module Three — Visualising Tool Calls and Plans

Lessons

• Showing Tool Invocations Without Exposing Internals
• Rendering Multi-Step Plans and Subagents
• Citations and Provenance in Agent Answers

Module Four — Approval and Interruption Flows

Lessons

• Designing Human-in-the-Loop Checkpoints
• Interrupting and Redirecting a Running Agent
• Error States and Recovery in Agent UIs

Module Five — Client State and Memory

Lessons

• Modelling Agent Sessions in Frontend State
• Persisting Conversations and Resuming Work
• Client-Side Context and Memory Controls

Module Six — Evaluating and Shipping Agent Interfaces

Lessons

• Testing Agent UIs with Recorded Trajectories
• Measuring Agent Interface Quality in Production
• Accessibility and Safety Review Before Launch