An AI teaching assistant for Process Dynamics and Control (apmonitor.com/pdc). Install it in your AI tool of choice and it becomes a course-aware TA: it knows the schedule, every assignment and TCLab activity, the simulation and control apps, the course's standard notation (FOPDT, IMC tuning, PID forms, the TCLab energy balance), and — most importantly — the course AI policy. It tutors, quizzes, and debugs with you; it does not do your graded work for you.

The same TA is packaged for Claude, ChatGPT / Codex, and Google Gemini. Pick your tool below.

1. Download pdc-course-ta.skill.
2. In Claude, open Settings → Capabilities → Skills and upload the file (or attach the .skill file to a chat — if your account allows skill creation, the file card shows a Save skill button).
3. Ask a course question in a new conversation; the skill triggers automatically, or invoke it with /pdc-course-ta.

Claude Code (terminal):

mkdir -p ~/.claude/skills && cd ~/.claude/skills
curl -LO https://github.com/APMonitor/pdc/raw/master/TA_Skill/pdc-course-ta.skill
unzip pdc-course-ta.skill && rm pdc-course-ta.skill

Custom GPT (needs a plan that includes GPT creation):

1. Go to Explore GPTs → Create (or chatgpt.com/gpts/editor) and open the Configure tab.
2. Name it PDC Course TA. Paste everything below the horizontal rule of pdc-course-ta-instructions.md into Instructions.
3. Under Knowledge, upload pdc-course-ta-knowledge.md.
4. Save (keep it private, or share the link with classmates per your instructor's guidance).

ChatGPT Projects (alternative, no GPT creation needed): create a Project, paste the same instructions into the project's custom instructions, and add the knowledge file to the project files.

Codex (the agentic coding tool / CLI): copy AGENTS.md and pdc-course-ta-knowledge.md into the folder where you run Codex — e.g., your course work folder or a clone of this repository (both files are at the repo's TA_Skill/; Codex picks up AGENTS.md automatically from the working directory tree).

Gemini Gem:

1. In Gemini, open Gems → New Gem (gem manager).
2. Name it PDC Course TA. Paste everything below the horizontal rule of pdc-course-ta-instructions.md into the Gem's instructions.
3. Add pdc-course-ta-knowledge.md as a knowledge file (if your Gemini version doesn't support knowledge uploads, paste its contents at the end of the instructions or into your first message).
4. Save and start chatting.

Gemini CLI: copy GEMINI.md and pdc-course-ta-knowledge.md into the folder where you run the CLI; it reads GEMINI.md automatically.

No special features available? Every tool falls back to the same trick: paste pdc-course-ta-instructions.md (below its horizontal rule) plus pdc-course-ta-knowledge.md as the first message of a conversation.

Talk to it like a TA at office hours. Examples of things it does well:

• Concept help — "Why does integral action remove offset but cause windup when the heater saturates?"
• Homework coaching — "I got Kp = 8.5 °C/% and θp = 140 s from my TCLab step test and my PI controller is sluggish — what am I doing wrong?" (It will sanity-check your fit and coach you to re-read the plot rather than just handing you numbers.)
• Quiz me — "Quiz me on second-order systems, one question at a time, and keep score." Every assignment's Generative AI Learning section has ready-made prompts like this.
• TCLab and code debugging — paste your Python error or controller code; it knows the tclab package, the digital twin (tclab.setup(connected=False)), and the velocity-form PID used in the course apps.
• Report coaching — bring your plots and numbers and it will critique your "What to Turn In" report against the assignment's questions.
• Project + EE support (classes 26–41) — circuits, sensors, actuator drive circuits, power budgets, SCADA architecture, and the electrical safety review for your microcontroller project.
• Going deeper — ask about Bode plots, state space, Smith predictors, or MPC; it will teach beyond the course and point you to follow-on resources like Dynamic Optimization.

The TA enforces the course AI policy in every tool, so expect it to redirect you if you ask it to:

• solve your specific graded assignment problem outright or write your report's results for you;
• answer quiz or exam questions being taken for a grade;
• invent data, plots, or "observations" for a lab you didn't run (it will point you to the TCLab digital twin or the browser apps instead — real data in minutes).

You are the engineer of record for everything you submit. Use the TA to learn faster, verify your work, and push further — then own the result.

All packages share the same three pieces of content: the TA instructions (policy + tutoring behavior), the course map, and the course notes. To adapt for your own course, edit those and re-export:

• Claude: the .skill file is a zip — extract, edit SKILL.md and references/*.md, re-zip, rename to .skill.
• ChatGPT / Gemini: edit pdc-course-ta-instructions.md (keep it under ~8,000 characters for GPT instruction boxes) and pdc-course-ta-knowledge.md, then re-paste/re-upload.
• Codex / Gemini CLI: edit AGENTS.md / GEMINI.md (they are identical) and the knowledge file.

pdc-course-ta-review.html shows the kind of behavioral testing used to check that the TA coaches instead of solving — worth re-running if you change the policy sections.