Overview

Create an instruction file that extends the dt-coach with the ability to generate descriptive image prompts during Method 5 (User Concepts). When users are developing concepts, visual representation helps communicate ideas that words alone struggle to capture. The coach produces structured prompts suitable for M365 Copilot image generation, descriptive enough for AI image tools to render meaningful concept visualizations while maintaining the lo-fi spirit of Design Thinking (concept sketches, not polished mockups).

Target File

.github/instructions/dt-image-prompts.instructions.md

Frontmatter

---
description: 'M365 Copilot image prompt generation for Design Thinking concept visualization — produces descriptive prompts for AI-assisted visual communication'
applyTo: '**/.copilot-tracking/dt/**/method-05*'
---

Required Content

Purpose and Scope

Image prompt generation supports Method 5 (Concepts) by helping users visualize their ideas. The coach offers to generate image prompts when:

• A concept is articulated but hard to describe verbally
• The user needs to communicate a concept to stakeholders
• Visual exploration might reveal aspects the verbal description missed

The generated prompts target M365 Copilot image generation capabilities, structured descriptive text that AI image tools can render into concept visualizations.

Prompt Generation Approach

The coach transforms concept descriptions into image prompts by:

1. Extracting the core idea and key visual elements from the concept
2. Adding context about the setting, users, and interactions
3. Specifying a visual style consistent with Design Thinking (sketch-like, conceptual, not photorealistic)
4. Structuring the prompt for M365 Copilot image generation format

Image Prompt Structure

Generated prompts follow this structure:

• Subject: What the image shows (the concept in action)
• Context: Where and when (the environment, stakeholders involved)
• Style: Sketch-like, whiteboard-style, or concept illustration, maintaining lo-fi DT spirit
• Focus: What aspect of the concept to emphasize
• Exclusions: Avoid photorealism, polished UI mockups, or final-product imagery

Lo-Fi Visual Alignment

Image prompts maintain Design Thinking's lo-fi principle:

• Request sketch or whiteboard aesthetics, not polished renders
• Emphasize concept communication over visual fidelity
• The image should invite feedback and iteration, not signal completion
• Coach guidance: "This visualization helps communicate the idea, it's a concept sketch, not a final design"

Integration with Method 5 Workflow

The coach offers image generation as an optional enhancement during concept articulation:

• Not every concept needs visualization, the coach suggests it when visual communication adds value
• Generated prompts are stored alongside concept artifacts in .copilot-tracking/dt/{project-slug}/method-05-concepts/
• Users copy generated prompts to M365 Copilot for image generation
• Generated images can be referenced in concept documentation but are not stored in the repository

Example Prompt Patterns

Include 2-3 example patterns showing how concepts translate to image prompts:

Token Budget

Target: ~800-1,200 tokens (on-demand tier, loaded only during Method 5 concept work)

How to Build This File

This is an .instructions.md file. Use the prompt-builder agent (not task-implementor) for the authoring phase.

Workflow: /task-research → /task-plan → /prompt-build → /task-review

Between each phase, use /clear to reset context, then attach the output artifact from the previous phase as input for the next.

Phase 1: Research

Research M365 Copilot image generation prompting and Method 5 integration.

Source Material: Image generation and DT concepts: #file:.github/instructions/prompt-builder.instructions.md for authoring standards, M365 Copilot image generation best practices, and the Method 5 concepts instruction file for integration points.

Steps:

1. Type /clear to start a fresh conversation.
2. Attach #file:.github/instructions/prompt-builder.instructions.md and #file:.github/instructions/dt-method-05-concepts.instructions.md (if available).
3. Copy the prompt below into chat and send.

/task-research topic="DT image prompt generation"

Research M365 Copilot image generation prompting for DT concept visualization.

Extract:
- M365 Copilot image generation prompt structure and best practices
- Method 5 (Concepts) workflow and integration points for visualization
- Lo-fi visual style descriptors (sketch, whiteboard, wireframe aesthetics)
- Prompt patterns that produce concept-level illustrations, not polished renders
- DT4HVE source materials on concept visualization techniques
- Prompt-builder compliance requirements for .instructions.md files

Output: DT image prompts research

Phase 2: Plan

Plan the instruction file, prompt generation approach and lo-fi alignment.

Steps:

1. Type /clear to reset the conversation.
2. Attach the research document from Phase 1.
3. Copy the prompt below into chat and send.

/task-plan

Plan the M365 Copilot image prompt generation instruction file.

Use the attached research document as input. The plan should cover:
- Purpose and scope (when to offer image generation during Method 5)
- Prompt generation approach (concept → visual elements → structured prompt)
- Image prompt structure (subject, context, style, focus, exclusions)
- Lo-fi visual alignment (sketch/whiteboard aesthetics, not photorealistic)
- Integration with Method 5 workflow (when to offer, where to store)
- 2-3 example prompt patterns
- Token budget (~800-1,200)

Output: .copilot-tracking/plans/{date}-dt-image-prompts-plan.md

Phase 3: Build

Author the instruction file using prompt-builder.

Steps:

1. Type /clear to reset the conversation.
2. Attach the plan document from Phase 2.
3. Copy the prompt below into chat and send.

/prompt-build

Build the M365 Copilot image prompt generation instruction following the attached plan.

Create .github/instructions/dt-image-prompts.instructions.md with:
- Frontmatter: description, applyTo targeting Method 5 artifact paths
- Purpose and scope — when the coach offers image generation
- Prompt generation approach transforming concepts into structured prompts
- Image prompt structure (subject, context, style, focus, exclusions)
- Lo-fi visual alignment — sketch/whiteboard, not photorealistic
- Integration with Method 5 workflow
- 2-3 example patterns showing concept-to-prompt transformation

Instruction style guides coach behavior rather than prescribing rigid templates.

Output: .github/instructions/dt-image-prompts.instructions.md

Phase 4: Review

Validate prompt generation produces useful results and maintains lo-fi alignment.

Steps:

1. Type /clear to reset the conversation.
2. Attach the plan document from Phase 2.
3. Copy the prompt below into chat and send.

/task-review

Review the image prompt generation instruction against the attached plan.

Validate:
- Frontmatter has description and applyTo targeting Method 5 paths
- Prompt generation approach clearly transforms concepts into structured prompts
- Image prompt structure defined (subject, context, style, focus, exclusions)
- Lo-fi visual alignment maintained — sketch/whiteboard, not polished renders
- Integration with Method 5 workflow documented
- 2-3 example prompt patterns included
- Token count within ~800-1,200 target
- Prompt-builder compliance verified

Output: .copilot-tracking/reviews/{date}-dt-image-prompts-review.md

After Review

• Pass — Open a PR with the instruction file.
• Iterate — Return to Phase 3 with the review document to fix identified issues.
• Escalate — Return to Phase 1 to investigate M365 Copilot image generation capabilities.

Authoring Standards

Follow .github/instructions/prompt-builder.instructions.md:

• applyTo targets Method 5 artifact paths (on-demand tier)
• Instruction style guides coach behavior rather than prescribing rigid templates
• Example patterns demonstrate the transformation from concept to image prompt

Success Criteria

• File created at .github/instructions/dt-image-prompts.instructions.md
• Frontmatter includes description and applyTo targeting Method 5 paths
• Prompt generation approach clearly transforms concepts into structured image prompts
• Image prompt structure defined (subject, context, style, focus, exclusions)
• Lo-fi visual alignment maintained, sketch/whiteboard aesthetics, not polished renders
• Integration with Method 5 workflow documented (when to offer, where to store, how to use)
• 2-3 example prompt patterns included
• Token count within ~800-1,200 target
• Passes task-reviewer validation against prompt-builder standards
• Artifact registered in collections/design-thinking.collection.yml with path and kind fields
• Artifact registered in collections/hve-core-all.collection.yml with path and kind fields
• npm run plugin:generate run after updating collection manifests