Overview

Create dt-rpi-review-context.instructions.md — a DT-aware instruction file that auto-loads when task-reviewer operates on DT artifacts. This redefines quality criteria so that the reviewer evaluates DT artifacts against Design Thinking standards (stakeholder coverage, synthesis depth, prototype fidelity, coaching effectiveness) rather than generic code review criteria.

This addresses a critical gap: without DT-specific review context, task-reviewer would apply standard code review criteria to artifacts that are coaching conversations, research notes, and design documents.

Directory context: The .copilot-tracking/dt/ path in applyTo is created per-project by the coach agent (#579 from Phase 2). See #583 for directory structure and handoff contract details.

Target File

.github/instructions/dt-rpi-review-context.instructions.md

Frontmatter

---
description: 'DT-aware task-reviewer context — quality criteria for Design Thinking artifacts'
applyTo: '**/.copilot-tracking/dt/**'
---

Required Content

Purpose

When task-reviewer is invoked for DT artifacts (instruction files, prompts, agent definitions, coaching state, method outputs), standard code quality criteria don't apply. This instruction file augments the reviewer's quality framework with DT-specific criteria.

DT Artifact Quality Criteria

Review Checklist Additions

When reviewing DT artifacts, task-reviewer adds these checks:

• Coaching tone: Language guides thinking, doesn't direct action
• Fidelity enforcement: Output quality matches the method's lo-fi/medium/high tier
• Stakeholder coverage: Multiple perspectives represented, not just the obvious ones
• Evidence grounding: Claims trace to research data (Method 2+) or acknowledged assumptions (Method 1)
• Iteration support: Content encourages revisiting and refining, not linear completion
• Token budget: Ambient ≤3,000; method ≤2,000; on-demand ≤3,000; industry ≤2,000

Anti-Patterns to Flag

• Directive coaching language ("You should...", "The correct approach is...")
• Production-quality output in early-method artifacts
• Missing stakeholder perspectives in synthesis
• HMW questions that are really solutions in disguise
• Handoff artifacts with leaked internal paths or planning IDs

Token Budget

Target: ~600-800 tokens (ambient tier — auto-loads on all DT artifact paths)

RPI Pipeline Workflow

1. task-researcher: Review existing task-reviewer patterns in HVE Core, understand how applyTo globs modify reviewer behavior, gather DT quality criteria from cumulative research.
2. task-planner: Plan the file — quality criteria table, review checklist additions, anti-pattern list.
3. task-implementor: Author following prompt-builder standards. Review criteria should augment (not replace) standard reviewer behavior.
4. task-reviewer: Self-validate — apply the DT review criteria to this instruction file itself as a test.

Success Criteria

• Instruction file created at .github/instructions/dt-rpi-review-context.instructions.md
• Frontmatter has applyTo: '**/.copilot-tracking/dt/**' glob
• Quality criteria defined per artifact type (coaching, method, deep, industry, handoff, agent, output)
• Review checklist additions cover coaching tone, fidelity, stakeholder coverage, evidence, iteration
• Anti-patterns listed for common DT artifact quality issues
• Token budget tiers documented
• Token count within ~600-800 target
• Passes task-reviewer validation against prompt-builder standards
• Each prompt, instructions, or agent file registered in collections/design-thinking.collection.yml with path and kind fields
• Each prompt, instructions, or agent file registered in collections/hve-core-all.collection.yml with path and kind fields
• npm run plugin:generate succeeds after collection manifest updates