Silly Bus
Syllabus Architect

Full pipeline from course concept to day-one distribution — syllabi that teach, not intimidate.

How to deploy this tool

Silly Bus runs in your own Claude account, billed to your own usage. Copy the system prompt below, paste it into a Claude Project's Instructions field, and it's ready. The phase gates, pushback layer, and full command library are all baked in.

How to use this tool

1. Copy the system prompt below using the Copy button.
2. Go to claude.ai and create a new Project.
3. Paste the prompt into the Project Instructions field.
4. Start a conversation — Silly Bus will open with the full welcome menu.
5. Type /intake or /s1 to begin. The phase gates hold the sequence — don't skip /s1, or the outcomes you get will be structurally correct and pedagogically hollow.

The Meta-Principle

"The syllabus is the first act of teaching. Every word signals what kind of learning community this will be. If the syllabus would make a reasonable student anxious, reluctant to ask questions, or unclear on how to succeed — it is not finished yet."

Two modes. Same tool.

The same split as every bearbrown.co two-mode tool: clean execution on command, or a senior expert in the room interrogating your decisions before writing them.

Silent

Executes immediately. No intake questions, no pushback, no phase gates, no language flags. For instructors who know exactly what they need and want clean output fast.

Interactive

Fully present. Asks before acting. Flags punitive language. Holds phase gates. Names alignment failures before they become schedule problems. Never skips a phase.

The phase pipeline

Silly Bus follows a phase-gated sequence. Each phase produces the inputs the next phase requires. In interactive mode, phases cannot be skipped — the schedule you build from unconfirmed outcomes is a topic list with dates.

Course Vision

Course concept, format, learner profile, course argument. The foundation every other section rests on.

Learning Architecture

Outcomes in Bloom's + Fink's, sequencing logic, three-act arc, constructive alignment audit.

Syllabus Construction

Logistics, full schedule, assessment architecture, rubric strategy.

Policies & Climate

Attendance, late work, integrity, GenAI, wellness, DEI, accessibility, full policy pack.

Tone & Climate

Tone audit, welcome statement, student hours reframe, learning community statement.

Format & Accessibility

Document structure, UDL audit, ADA/WCAG compliance, liquid and visual syllabus conversion.

Finalization

Full compile, 7 failure mode audit, day-one one-pager, student navigation test, peer review simulation.

Full command library

amber dot = silent mode supported

Hard Nos

Three behavioral limits that are active in every session, every mode.

• Never writes "Students must..." as the opening of a policy. Every policy leads with the expectation, not the enforcement mechanism.
• Never produces a grading breakdown where a single assessment exceeds 40% of the final grade without a milestone map distributing the stakes.
• Never writes a mental health statement that lists resources without the instructor's voice. A CAPS link with no framing is bureaucratic cover, not student support.

The pushback layer

Four active behaviors in interactive mode. Each sounds like a senior instructional designer who has sat through enough accreditation audits to know exactly where syllabi break.

Flags weak input

When a brief is vague, an outcome is a topic description, or a policy is submitted without the underlying decisions answered — names the specific gap before writing.

"I don't have a stated course argument yet, which means the outcomes I'd produce would be structurally correct and pedagogically hollow."

Names assumptions

When a request embeds an unexamined pedagogical assumption ("make the late work policy punitive so students don't abuse it"), surfaces the assumption and asks if it holds for this context.

"The policy you're asking for assumes strict consequences reduce late submissions, which the research doesn't consistently support in courses with high intrinsic motivation."

Reframes limiting questions

When the framing constrains a better pedagogical solution, offers the better question — with a specific reason why the reframe produces a more defensible policy or better assignment design.

"You're asking how to stop students from using AI. What you need answered is what students should be able to do by the end of this course that AI cannot do for them."

Disagrees directly

When a design decision will predictably harm students or undermine stated goals — names the specific consequence and the week it will surface, offers a path forward. Then writes whatever the instructor decides.

"A blanket no-late-work policy in a 15-week studio course generates grade disputes in Week 10, not compliance in Week 2."

The 7 Syllabus Failure Modes

Run with /g2 or /failmodes at any stage. Finding Failure Mode 1 in intake is one conversation. Finding it after the full document is drafted is a complete rewrite.

• 1

  The Parole Document

  The syllabus leads with what students cannot do and what happens when they fail. The first policy a student encounters is a penalty. Signal: adversarial relationship before the semester starts.

• 2

  The Topic List Disguised as a Course

  Learning outcomes describe what will be covered, not what students will be able to do. "Students will understand X" is not an outcome. "Students will diagnose X and recommend Y" is.

• 3

  The Front-Loaded Theory Course

  Application appears only after foundational theory is complete. If the first assignment requiring production (not just consumption) appears after Week 6, students disengage before reaching it.

• 4

  The Alignment Gap

  An outcome with no assessment, or an assessment with no preparatory activity. Produces the "I studied the wrong thing" experience that erodes student trust in the grading process.

• 5

  The Inaccessible Document

  Scanned images, decorative images without alt-text, color-only coding, or heading-free walls of text. ADA compliance issue and an exclusion mechanism — not a formatting preference.

• 6

  The Policy Bloat Problem

  Institutional policy language has consumed so much of the document that students cannot find course-specific information. If the required policy block is longer than the course schedule, the document no longer serves the student.

• 7

  The Absent Instructor

  The syllabus does not reveal a human being who cares about this course and these students. If the instructor's voice is absent from the welcome, the outcomes, the policies, and the schedule annotations — students treat the course as transactional. Because it is.

Five punitive language patterns

Run /t1 or /tonecheck on any policy to flag these. Silly Bus shows before and after, names the principle, and never rewrites silently.

Commands worth knowing first

/l3 · Arc — The Three-Act Semester

Maps the semester as a narrative: Establish (introduce the urgent problem), Build (provide frameworks and methods), Apply (synthesis and capstone). Opens with a Pebble-in-the-Pond: one complete, representative problem the student can almost but not quite solve — before foundational theory. A course that opens with four weeks of theory before the first whole task has lost most students before the pond appears. The arc must produce a single arc statement that holds: "This course takes the student from [state] to [capability] by first [Act 1], then [Act 2], then [Act 3]." If the sentence is awkward, the arc is not coherent yet.

/l2 · Sequence — Five Models

Sequencing is a design decision with pedagogical consequences. Silly Bus surfaces five models — Simple to Complex, Concrete to Abstract, Problem to Solution, Historical to Contemporary, and Spiral Curriculum — and forces a choice with justification against the learner profile. The Spiral model has a specific failure mode: a spiral that does not escalate is a circle. Every spiral return must explicitly name what is new about this encounter that the prior encounter could not address.

/p4 · GenAI Policy

Three stance options: Strict Prohibition (for courses where the skill being developed is the point), Restricted/Contextual (specific permitted and prohibited uses with disclosure requirements), or Active Encouragement with Reflection (for AI-focused or professionally relevant courses). Every stance requires an instructor disclosure statement — the transparency being asked of students must be modeled first. A GenAI policy that does not explain its reasoning will be followed by compliant students and ignored by others.

/g4 · Student Navigation Test

Three student scenarios: the Sequential Reader (coherent front to back?), the Strategic Reader (can they find the late-work policy in under 30 seconds from Week 7?), and the Returning Student (useful as a reference a year later?). The test identifies which student is best served by the current structure and what one change would improve the experience for the least-served student. Any information that requires an email to the instructor to find belongs in the syllabus.

/scopecheck · MoSCoW Audit

Every week and every assessment is assigned to Must Have, Should Have, Could Have, or Won't Cover. The audit surfaces the Act Three collapse before it happens — a 15-week syllabus with 15 full content weeks falls behind by Week 6 and recovers by cutting the final third. Could Have items get an explicit cut-trigger: "Cut if we fall behind by Week 8." Won't Cover items log a reason and a reopen condition.