# Design Harness Roadmap

## Definition

The Design Harness is the future downloadable version of Agentic Design Ops: a folder of context files, prompts, templates, checklists, and examples that a designer can copy into a project and adapt.

It should feel like an operating system for working with AI agents, not a prompt pack.

The Harness should be introduced early as a light map, but the full setup should come after the foundations. The learning sequence is:

1. AI design process
2. light Harness map
3. skill format
4. project memory
5. design context
6. orchestration
7. source trust and QA
8. full Harness setup

## Product Promise

Help designers make agents more useful by giving them:

- project context
- design-system rules
- product constraints
- source-trust rules
- critique loops
- validation checklists
- reusable task patterns
- examples from real product work

## V0

The website explains the concept and collects the first workflows.

Deliverables:

- foundation docs
- sitemap
- workflow placeholders
- harness page
- reference page
- two article templates
- source/content inventory
- editorial roadmap

V0 source material:

- current `AGENTS.md`
- current `DESIGN.md`
- `docs/design-system/*`
- `docs/CONTENT-INVENTORY.md`
- `docs/EDITORIAL-ROADMAP.md`

## V1

A downloadable folder that designers can copy into a project.

Possible contents:

- `AGENTS.md`
- `DESIGN.md`
- `PRODUCT.md`
- `VOICE.md`
- `SOURCE-TRUST.md`
- `QA-CHECKLIST.md`
- `ARTICLE-TEMPLATES.md`
- `prompts/`
- `templates/`
- `examples/`

Suggested V1 folder structure:

```text
design-harness/
  AGENTS.md
  DESIGN.md
  PRODUCT.md
  VOICE.md
  SOURCE-TRUST.md
  QA-CHECKLIST.md
  prompts/
    design-critique.md
    source-check.md
    html-mockup.md
    article-draft.md
  templates/
    workflow-guide.md
    editorial-essay.md
    case-study.md
    decision-doc.md
  examples/
    planark/
```

V1 should be built from:

- `article-template.html`
- `editorial-template.html`
- `planark-build-book/03-html-mockups.md`
- `planark-build-book/05-debugging-evidence.md`
- `planark-product-book/05-real-data.md`

## V2

A guided setup version.

Possible contents:

- setup checklist
- project questionnaire
- prompt generator
- context-file generator
- example project
- simple "choose your stack" guide
- starter content audit
- recommended first workflows

## V3

A maintained public toolkit.

Possible contents:

- versioned releases
- install instructions
- changelog
- community examples
- optional CLI or web generator

## Modules

### Context Module

Files:

- `AGENTS.md`
- `DESIGN.md`
- `PRODUCT.md`
- `VOICE.md`

Purpose: make the agent understand the product, audience, design language, and working rules before it edits.

### Workflow Module

Files:

- workflow prompts
- task checklists
- review loops

Purpose: turn repeated work into named processes.

### Validation Module

Files:

- `SOURCE-TRUST.md`
- `QA-CHECKLIST.md`
- visual QA checklist
- source/log tracing checklist

Purpose: prevent plausible AI output from passing as finished work.

### Publishing Module

Files:

- article templates
- LinkedIn adaptation template
- editorial roadmap template

Purpose: turn build notes into public writing without losing proof.

## Open Decisions

- Should the first download be a `.zip`, a GitHub repo, or a copyable folder on the site?
- Should the harness be generic first or Planark/example-led first?
- How much of the PM skills pack pattern belongs inside the first design harness?