Every agent has a defined scope. They only do their part, then hand off
to the next one. Here is who does what.
Keeps the whole pipeline moving. Creates tasks, assigns work to other agents, enforces approval gates, and makes sure nothing gets skipped or stuck.
- Launches new course lanes and stages child tasks
- Monitors progress and unblocks downstream agents
- Packages decision documents for team review
- Enforces work-in-progress limits
Manages all login sessions for Canvas and MindTap. Makes sure every other agent has working access before they start. If a session expires, this agent refreshes it.
- Bootstraps Canvas and Cengage/MindTap authentication
- Publishes readiness status for each course
- Recovers from session failures automatically
Reads the existing Canvas master course from top to bottom. Extracts the syllabus, module structure, assignments, grading rules, and learning outcomes. Strictly read-only.
- Pulls course structure, outcomes, and grading logic
- Identifies student-facing clarity issues
- Documents what the course currently looks like
Maps the entire MindTap course: every activity, reading, quiz, and exercise. Classifies content types, extracts textbook source material, and documents what is auto-graded vs. manually graded.
- Full structural map of the MindTap course shell
- Textbook source-text extraction for coverage analysis
- Identifies essay questions that need MC replacement
The heart of the operation. Compares Canvas and MindTap evidence, finds gaps in alignment, and produces the instructional design package: alignment matrix, weekly blueprint, assessment plan, and student experience review.
- Alignment matrix mapping outcomes to content to assessment
- Week-by-week instructional blueprint
- Assessment blueprint with Bloom's taxonomy targets
- Student experience review scored across 7 dimensions
Executes the approved rebuild plan in a MindTap duplicate copy. Reorganizes the course structure, applies naming standards, configures quiz and homework settings, and sets point values. Every change is logged with before-and-after evidence.
- Structure normalization (folders, titles, ordering)
- Template and settings configuration
- Full audit trail of every modification
Creates and replaces assessment items following the approved blueprint. Writes multiple-choice questions with CHCP-specific workplace scenarios, replaces essay questions where needed, and maps every item to a learning objective.
- New MC item creation via Learnosity
- Essay-to-MC replacement with Bloom's level preservation
- Distractor design targeting real misconceptions
Runs an 82-point instructional quality scorecard on the rebuilt course. Checks alignment, scaffolding, assessment quality, student experience, accessibility, faculty usability, and operational trust. Courses that do not pass go back for revision.
- Technical QA: Does everything load and link correctly?
- Semantic QA: Is the content accurate and on-topic?
- Instructional QA: Would students actually learn from this?
Produces the instructor handoff document when a course is complete. Includes weekly teaching maps, known student trouble spots, grading notes, and the reasoning behind every design decision. This is what faculty receive.
- Weekly teaching guide with pacing notes
- Known friction points and how to address them
- Design rationale for instructors