Designing a Teacher's Work with AI
Learn why teachers build their own AI skills and apps, and how roles, materials, and goals guide the work.
Vibe
3 seriesHands-on AI Agent Use for Teachers
Learn Your VS Code Workspace for Working with AI
Follow VS Code demos to open a folder, write AI work instructions, save a file, and check its Markdown preview.
Activity 1: Create a Skill — Read an Existing Worksheet
Separate what AI actually reads from what it infers, add your teaching intentions, and save a reviewed Markdown analysis of an existing worksheet.
Activity 1: Create a Skill — Write a New Worksheet
Choose changes to a familiar worksheet, review and approve a plan, then draft and separately review student materials and teacher answers.
Activity 1: Create a Skill — Make a Printable PDF
Lay out a reviewed fractions worksheet with print-oriented Paper CSS and KaTeX, revise its browser print preview, and save an A4 PDF.
Activity 1: Create a Skill — Save the Procedure as a Skill
Extract a reusable skill from verified source analysis, reviewed manuscripts, and A4 print checks, then apply it to different materials.
Activity 2: Web Automation — Prepare for Practice
Create a workspace separate from worksheets and decide what to delegate for document searches and pending-approval checks.
Activity 2: Web Automation — Connect AI to Chrome
Prepare Chrome separately from everyday browsing, ask AI to read a public page, and close the connection after practice.
Activity 2: Web Automation — Find Official Documents
Ask for documents in the work screen you logged into, clarify the search together, and compare one result yourself.
Activity 2: Web Automation — Get Pending-Approval Notifications
Check a pending list with AI, then optionally set up computer notifications. Follow the activity from a visible test alert to stopping it.
OpenCode
[OpenCode Field Notes Part 1] Why did work get harder even with more tools?
A personal account of fragmented setup, long-session context fatigue, and why I moved to a more sustainable CLI workflow.
[OpenCode Field Notes Part 5] DCP context optimization and daily operating routine
How I used DCP to keep long sessions stable, reduce context fatigue, and sustain practical coding loops.
Pi
[Pi Series 1] Why I started looking at pi instead of heavier coding agents
This post explains why LLM coding agents can feel heavy, and why the lightweight terminal coding harness pi is worth considering.
[Pi Series 2] Install pi and start your first session
Install pi, connect a model through login or an API key, and start your first session inside a project folder.
[Pi Series 3] Manage sessions and project context
Learn how to continue and branch pi sessions, and how to pass project rules through AGENTS.md and CLAUDE.md.