FCC for Students

Welcome to the student track. This set of pages is written for university-level learners — undergraduate and early-graduate — in computer science, information science, data science, human-computer interaction, public-policy, and adjacent fields. The material is also usable by instructors designing assignments or capstone projects that incorporate FCC.

What you will learn

By working through this track you will:

• Install FCC in a classroom-friendly environment and run your first scenario.
• Read and write persona YAML using the R.I.S.C.E.A.R. spec.
• Design a research project using the FCC workflow as instrument.
• Participate in a persona role-play exercise that clarifies agent-system reasoning.
• Apply evaluation rubrics to your own outputs.
• Produce submission-ready assignment artifacts (lab reports, posters, papers).

How this track is organized

Page	Focus	Approx. time
Getting Started	Install + first run	45 min
Assignment Patterns	Eight reusable assignment shapes	reading
Research Project Workflow	Plan-build-write for a term paper	2-3 hours
R.I.S.C.E.A.R. Exercise	Hands-on persona authoring	60 min
Persona Role-Play	Classroom activity	45 min
Evaluation Rubrics	Grade-ready rubrics	reference

Prerequisites

• Comfort installing Python packages (pip install ...).
• Familiarity with YAML (you have probably edited GitHub Actions files or pyproject.toml).
• A text editor (VS Code or similar) and basic Git.
• Willingness to read specification documents.

You do not need prior AI or ML experience. The mock mode of FCC produces deterministic outputs without calling any AI API, so you can complete the entire track on a laptop, offline, with no accounts, no credits, and no privacy worries.

Why FCC for a student project?

Three reasons FCC is unusually well-suited to coursework:

1. Inspectable outputs. Every run produces a structured trace. You do not need to "peel apart an LLM call" to analyze behaviour — it is already structured.
2. Deterministic mode. For reproducibility of grade-relevant experiments, set provider: mock and results are bit-identical across students and grading sessions.
3. Low ceiling, high ceiling. A three-persona "hello world" takes 30 minutes. A capstone-scale project (build a vertical vocabulary for a new domain) takes a full semester. Both use the same APIs.

Course integration scenarios

FCC has been successfully used in:

• Software engineering courses — as the target system for a software architecture assignment. Students read the ADRs and evaluate the design choices.
• AI ethics courses — the compliance module is a direct vehicle for studying EU AI Act / NIST AI RMF. Students audit an FCC deployment and write a policy brief.
• HCI courses — the collaboration module (human-in-the-loop sessions, approval gates, scoring) is a living example of mixed-initiative design.
• Information-science courses — the knowledge graph + RAG modules are a compact, instructable example of linked data in practice.
• Capstone projects — build a new constellation-vertical (e.g., a vocabulary for a niche industry the student cares about).

What this track is not

• Not a general Python tutorial. We assume you can install packages and run scripts.
• Not a general LLM course. FCC is usable without any LLM. Using hosted LLMs with FCC is an advanced path covered in the "intermediate" track elsewhere in docs/tutorials/.
• Not a walkthrough of every feature. FCC has 164 personas and 7 workflow graphs. We cover a minimum viable subset.

Ethical and honour-code guidance

Several instructors have asked for explicit guidance on how to use FCC in assignments without crossing academic-integrity lines. Here is our suggestion:

• Cite FCC when you use it. A standard citation template is provided in Research Project Workflow.
• Distinguish what FCC produced from what you produced. In lab reports, use separate sections or explicit callouts.
• For graded exercises, use provider: mock unless your instructor explicitly approves using a hosted LLM. This prevents inadvertent plagiarism of generated text and keeps results reproducible for grading.
• Share scenarios, not completed outputs. When collaborating with peers, swap scenario JSON files. Do not swap trace outputs.

Related material

• Archetype atlas — if you are designing your own persona, start here.
• ADRs — great source of short reading assignments in a software-engineering course.
• Guidebook chapters 1-3 — deeper foundations for a full-semester module.

Where next

Start with Getting Started. If you are an instructor and already have a course topic, jump to Assignment Patterns to find a ready-made assignment shape.