MuggsOf
Computer Science
Two terms, one craft.
specify · generate · judge
Learning to say what you want, read what the machine makes of it, and tell the difference.
Helpful links
▸Computer Science I — identity charter & AP CSA on-ramp
A governance document. 2026–27 · MuggsOfCompSci · v1.1
How to use this: Every activity spec for Computer Science I must reference this file. Any activity Claude Code (or anyone) builds for CS I must pass the Identity Test in §4. If an activity fails it, the activity is wrong — not this charter. When the AP on-ramp and the CS I identity appear to conflict, the identity wins and the on-ramp yields.
1. What Computer Science I is (the non-negotiable identity)
Computer Science I is a course about the power of AI to let a person understand and wield technologies far beyond what they could build by hand, and to create genuinely useful things.
Its student lives in the builder-not-consumer world: they direct AI, read what it made, judge it, and iterate — as the author of the work, not its user.
The promise to students, kept literally:
You will not be expected to write a line of code from scratch, if you don't want to.
This promise is not in tension with anything below. Directing, reading, tracing, judging, decomposing, and specifying are not writing code from scratch — they are the literacy of a person who commands code. The entire course can be taught without a student ever hand-authoring a for loop.
2. The relationship to AP CSA (support, not purpose)
CS I supports the spring AP CSA course, but support is a byproduct, never the point.
- The real work of a CS I activity is always building/understanding something with AI.
- That this work also seeds AP CSA-relevant thinking (the object model, tracing fluency, decomposition) is a quiet consequence — invisible to students, incidental to the lesson.
- The seeding lives entirely on the reading-and-judging side — which is exactly the builder posture. Nothing that helps AP CSA requires scratch-coding.
The line that must never be crossed: an activity may never exist in order to prep AP CSA. It exists to build something real with AI. If it happens to strengthen a spring skill, good — but that can never be its stated reason for being.
The best on-ramp isn't front-loaded syntax; it's a student who arrives in spring motivated, unafraid of complexity, and convinced they are powerful. That disposition outperforms a head start on Java.
3. What CS I quietly seeds (margin note, not curriculum)
Recorded here so the benefit is known — but it must stay a margin note in any activity, never the headline.
| Builder skill the activity foregrounds | Quietly also seeds (AP CSA) |
|---|---|
| Reading & judging AI-generated code | Unit 1 object model · tracing (→ 55% MCQ) |
| Writing a clear spec before generating | decomposition · class-design intuition (FRQ 2) |
| Predicting what code will do, then checking | control-flow reasoning (Unit 2) |
| Co-authoring the AI Policy | data-collection ethics (CED 4.1) · responsible use |
If an activity's primary framing is the right-hand column, it has failed §4.
4. The Identity Test
Scope. This test governs student-facing learning activities — labs, builds, lessons, projects. It does not govern assessments (pretest, Q1/Q2 midterms, post-test, unit assessments), which exist to measure and answer to the assessment spine in the Roadmap. An activity that happens to produce assessable evidence is still an activity. An instrument administered to produce a score is not.
The test has two tiers. Gates are few, falsifiable, and carry veto power. Commitments describe how the course should feel and are reviewed at the module level.
4.1 Gates — an activity failing either gate is reworked
- AP-support is silent. Any AP CSA benefit is a consequence. It is unstated to students and is never the activity's reason for being. If the activity's purpose cannot be stated without referring to spring, it fails.
- No forced scratch-coding. No student is required to hand-write code from scratch. They may choose to; they are never expected to.
Two gates, not six. Each names a specific failure mode this course was built to prevent, and each can be checked by someone other than the author.
4.2 Design commitments — reviewed per module, not per activity
- Builder-first. The purpose is to build, understand, or wield something real — not to prepare for a test. Understanding the machine counts: the Systems, Networks, and Data buckets are builder work even where no model is involved.
- Reading and judging over hand-authoring. Where code is involved, the student's core work is directing, reading, tracing, and judging what the machine produced. This is a claim about code, not about all cognition. Activities that ask students to create instruments, visualizations, data dictionaries, or designs are authoring by intent and are in scope.
- Real and useful. It aims at something a student would find genuinely interesting or useful, not a contrived exercise.
- Awe-preserving. It protects the sense that AI makes the student powerful and complex things reachable. It does not reduce the course to drill.
A single activity may sit lightly against one of these. A module that sits lightly against the same one throughout is a design failure and gets reworked.
4.3 Revising this test
If an activity fails a gate, rework the activity.
If a gate repeatedly rejects work that plainly belongs in this course, that is evidence about the gate, not about the work. Amend it deliberately — in writing, with the reason and the date, in this file. Charter revision is a considered act, not a convenience. But a charter that forecloses its own revision does not stay fixed; it drifts silently instead, which is worse.
5. Spec boilerplate
Paste into every CS I activity spec.
This activity is for Computer Science I and is governed by
CS1_TO_APCSA_ONRAMP.md(the Identity Charter). Before finalizing, verify it clears both §4.1 gates: AP-support is silent, and no scratch-coding is required. Check it against the §4.2 design commitments and name any it sits lightly against — do not drop them silently. CS I identity: AI as the power to understand and build beyond hand-coding; builder-not-consumer; no student is expected to write a line of code from scratch. AP CSA support is a silent byproduct, never the activity's purpose. If identity and on-ramp conflict, identity wins.
6. Why the front-load doesn't threaten this (the reassurance)
Concepts are permanent; syntax is perishable. CS I banks the thinking (reading, tracing, decomposition) through genuine AI-building; AP CSA later attaches Java's syntax to concepts students already own. The CED itself endorses this order — its Unit 1 guidance says to study, explain, and hand-trace existing code before writing from scratch, which is exactly the SDD read-and-judge method CS I is built on. So the on-ramp is served precisely by keeping CS I fully itself — a pure AI-building course — not by diluting it into pre-AP.
In one line
CS I is the course about the power of AI to build beyond your own hands; the AP on-ramp rides underneath in silence, and if the two ever pull against each other, CS I's identity wins.