| name | ai-agent-design-spec |
| description | Turn dissertation evidence into an AI agent design specification for supporting active learning implementation, including user scenarios, functions, interaction flow, boundaries, failure modes, and design rationale. |
AI Agent Design Spec
Use when the user asks to design, refine, evaluate, or document the AI agent proposed in the dissertation.
Inputs
Look for:
DISSERTATION_BRIEF.md
- teacher interview findings, concept-card responses, design-elicitation outputs, or confirmed co-design outputs
- teacher interview/workshop findings
research-wiki/DESIGN_RATIONALE.md
research-wiki/FINDINGS_INDEX.md
- chapter drafts about design implications
Workflow
- Identify the teaching problem the agent supports.
- Define primary users:
- university teachers
- teaching assistants or learning designers, if relevant
- students only if the dissertation explicitly includes them
- Map teacher needs and concerns to agent capabilities.
- Specify:
- user scenarios
- agent goals and non-goals
- core functions
- interaction flow
- inputs and outputs
- human control points
- failure modes and recovery
- evidence source for each design decision
- Separate evidence-backed requirements from speculative features.
- Produce a design rationale that links back to active learning and adoption conditions.
Output
Create or update:
design-specs/AI_AGENT_DESIGN_SPEC.md
design-specs/USER_SCENARIOS.md
design-specs/DESIGN_RATIONALE.md
Required Sections
Use this structure unless the user asks otherwise:
- Purpose
- Intended users
- Teaching scenarios
- Core capabilities
- Boundaries and non-goals
- Interaction flow
- Teacher control and override points
- Data and privacy assumptions
- Failure modes
- Evidence-to-design traceability table
Guardrails
Read ../dissertation-shared/references/privacy-and-ethics.md when participant data is involved.
Do not claim the agent is effective unless evaluation evidence exists.