Skip to content

Assignment Patterns

Eight reusable assignment shapes that instructors and students have used successfully with FCC. Each pattern specifies the learning outcome, the artifact, the approximate effort, and the grading anchor.

Pattern 1: Read-and-summarize an ADR

Learning outcome: understand the reasoning behind one architectural choice.

Artifact: a 500-800 word summary that (a) restates the decision, (b) identifies one alternative that was rejected and explains why, (c) proposes one test that would reveal if the decision was wrong.

Effort: 2-3 hours.

Grading anchor: students who merely paraphrase the ADR score C. Students who articulate a rejected alternative and a falsifying test score A.

Pattern 2: Persona design

Learning outcome: apply the R.I.S.C.E.A.R. specification to a domain the student chooses.

Artifact: one YAML persona file that passes fcc personas validate, plus a 300-word design memo explaining the choices made for each of the ten R.I.S.C.E.A.R. fields.

Effort: 3-4 hours.

Grading anchor: whether the constraints meaningfully narrow behaviour; whether responsibilities are actionable; whether the collaborator list is plausible.

Pattern 3: Scenario run + trace analysis

Learning outcome: interpret a structured trace and draw conclusions.

Artifact: one scenario JSON, one trace JSON, and a 600-word analysis identifying the longest persona-to-persona interaction chain, the action type most frequently executed, and one opportunity to reduce redundant work.

Effort: 3-5 hours.

Grading anchor: quantitative claims must cite trace fields; qualitative claims must reference specific persona IDs.

Pattern 4: Compliance audit of a sample deployment

Learning outcome: map a technical artifact to a regulatory requirement.

Artifact: a one-page policy brief (see the for-policy-makers track for the template) following fcc compliance-audit on a provided scenario.

Effort: 4-6 hours.

Grading anchor: whether every claim in the brief is backed by an evidence-item ID; whether the EU AI Act / NIST RMF crosswalk is correctly applied.

Pattern 5: Vertical-vocabulary extension

Learning outcome: scale the persona model into a new domain.

Artifact: 3-5 new personas in a new category; a vocabulary-provider plugin adapter; a short README for the extension.

Effort: 8-12 hours (small-group project, 2-3 weeks).

Grading anchor: whether the vocabulary is internally consistent; whether the plugin adapter follows the documented contract (see ADR-0001 Plugin System).

Pattern 6: Knowledge-graph construction

Learning outcome: represent a small domain as a knowledge graph using FCC's 9 node and 9 edge types.

Artifact: a .jsonld knowledge graph for a domain of the student's choice, with at least 30 nodes; plus a short analytical report using three SPARQL-style queries to answer non-trivial questions about the graph.

Effort: 6-8 hours.

Grading anchor: whether the graph is well-typed; whether the queries produce non-obvious findings.

Pattern 7: Capstone — build a small constellation

Learning outcome: end-to-end system design, versioned software, documentation.

Artifact: a small constellation-style repo that consumes FCC vocabulary, ships 5-10 personas, has a Mermaid architecture diagram, passes CI, and publishes a short paper or poster.

Effort: full semester, 10-15 hours/week.

Grading anchor: is the system self-describing enough that a fresh reader can onboard in 15 minutes?

Pattern 8: Replication study

Learning outcome: practice reproducibility.

Artifact: a replication report of a published FCC example (from notebooks/ or demos/), including any deviations observed, attempts to close the gap, and a reflection on what reproducibility means in practice.

Effort: 6-10 hours.

Grading anchor: honesty about gaps; depth of hypothesis for each deviation.

Using these patterns in a course

A single-course can absorb 2-3 of these patterns comfortably. Recommended combinations:

  • Intro software engineering (one semester): Pattern 1 (week 2), Pattern 3 (week 6), Pattern 5 (group project weeks 9-15).
  • AI ethics (one semester): Pattern 2 (week 3), Pattern 4 (week 7), Pattern 8 (final project).
  • HCI / interaction design (one semester): Pattern 2 (week 4), Pattern 6 (week 10), persona role-play (see companion page) as mid-semester exercise.

Group work

Patterns 5, 6, 7 scale well to groups of 2-4. Patterns 1, 2, 4 are better as solo work. Pattern 3 is borderline — pairs work well if they divide "run it" from "analyze it."

Academic-integrity notes

  • Submit the scenario JSON and trace JSON along with the write-up. These are the reproducible core.
  • If you used a non-mock provider, disclose the provider, model name, and date in an appendix.
  • If two students produce byte-identical traces with the same scenario + mock provider, that is expected and fine.

Where next

Ready to pick one? Go to Research Project Workflow for the project-plan template. Want to deepen persona authoring first? Go to R.I.S.C.E.A.R. Exercise. Running a classroom activity? See Persona Role-Play.