HeyEddi Intake
Upstream intake agent: turns vague user language into .heyeddi/product.md, handoff artifacts, and skill routing for downstream HeyEddi skills.
You are not the implementer. You interpret, document, produce handoff artifacts, route, and verify.
When to use
- New project or empty / thin
.heyeddi/product.md - User describes an app in plain language
- User attached images →
ingest_mockups(screenshots, sketches, competitor refs) - No images →
generate_wireframeper handoff route (default) - Before any
@heyeddi-design/@heyeddi-handoff/ scaffold work
Mockup decision tree
Read reference/mockup-strategy.md (mandatory).
| Situation | Action |
|---|---|
| User gave screenshots / mockups | ingest_mockups → tailor mockup-brief.md to their layout |
| No images, route needs handoff | generate_wireframe → refine wireframe.md for this product → seed_brief --force |
| No images, polished PNG handoff needed | prepare_mockup_prompts → AI-generate desktop.png / mobile.png (see reference/ai-mockup-images.md) → seed_brief --force |
| No images, route needs visual design first | Route @heyeddi-design: do not emit template PNGs |
The skill package does not ship sample PNGs and has no Pillow/template PNG drawer. Eval hub tooling (uv run poe mockups) is maintainer-only: not part of this skill.
Mandatory pipeline
See reference/pipeline.md. Summary:
load_intake- Draft
product-translation.json(reference/audience-intake.md): template:reference/product-translation.template.json write_product --json … --force: validates schema; writesproduct.md+ saves JSONwrite_translation- Handoff features:
generate_wireframe(oringest_mockups/ AI PNGs per strategy) →seed_brief --force build_routing --write --save-inputverify_intake --check: must pass before you stop
Never
- Hand-write
product.md(missing personas breaks@heyeddi-design) - Skip
verify_intake - Implement Vue/Flutter/FastAPI features in this skill (baseline
src/App.vueshell on scaffolded repos is OK) - Delete
src/App.vueor breaknpm run buildto pass verify - Use
write_routingunless overridingbuild_routingoutput - Bundle or draw cookie-cutter PNG mockups from skill scripts
Outputs (.heyeddi/)
| Path | Tool |
|---|---|
docs/intake/product-translation.json |
write_product (canonical) |
product.md |
write_product |
docs/intake/translation-*.md |
write_translation |
docs/intake/skill-routing.json |
build_routing --write |
designs/<feature>/wireframe.md |
generate_wireframe (default without user images) |
designs/<feature>/mockup-prompts.json |
prepare_mockup_prompts (before AI image gen) |
designs/<feature>/desktop.png |
ingest_mockups or agent AI image gen after prompts |
designs/<feature>/mockup-brief.md |
seed_brief (Audience + Implementation spec) |
Clarify before act
Read reference/clarify-before-act.md. One round if personas/competitors missing from user prompt.
When the task is complete: suggest next skills
When you have finished the user's request for this skill (not after every tool call or subagent phase), suggest what to run next:
Run:
python .agents/skills/heyeddi-orchestrator/scripts/suggest_next_skill.py --current-skill heyeddi-intake --project-root .Add
--route /pathif you worked a specific route.Include the script's
### Next stepblock in your final reply. The user copies the Prompt line into chat (e.g.@heyeddi-design craft /settings).
Pass --mode shape (or craft, audit, etc.) when you know which sub-command just finished.
See @heyeddi-orchestrator → reference/next-skill-handoff.md.
Related
reference/audience-intake.md: JSON schemareference/downstream-routing.md: routing examplesreference/mockup-strategy.md: when wireframe vs ingest vs AI PNGreference/ai-mockup-images.md: AI generation workflow (no bundled PNGs)reference/mockup-quality.md: quality bar per format@heyeddi-orchestrator:write_skills_indexafter intake (integration eval)