# Studio product definition

## Purpose
Studio is a personal and AI reference for making interfaces, voice experiences, videos and images. It gathers useful tools, techniques, recipes, examples and principles. Its job is to help a reader choose what to use, understand why and find the practical next step.

## Confirmed requirements
- Use the approved visual-library direction A: a light interface, left navigation, compact search and illustrated cards with captions below the image.
- Organise by UI, Voice, Video and Images, with cross-cutting access to Workflows, References and Principles.
- Give resource links a meaningful preview. Use an actual existing capture where its identity is known; otherwise use an explicitly labelled capability map. Never present the map as a screenshot or tested output.
- Distinguish engines, applications, wrappers, editors, sources and recipes. Keep one canonical product record with multiple domain memberships when appropriate.
- Compare alternatives doing the same task. Explain task fit with stars, rationale, inputs, outputs, limitations and a suggested route. Unrated tools remain visible.
- Serve human browsing and AI retrieval from the same catalogue, without putting technical implementation details into the main browsing flow.

## Decision and evidence rules
Stars are editorial task suitability and operating-effort judgments, not measured output-quality rankings. There is no overall average. A catalogue review date is not proof of a fresh source visit, funded account or successful runtime test. Each record retains its evidence note and primary-source links when available. Account and private runtime details remain outside the public catalogue.

## Preservation boundary
Generated catalogue pages supplement the existing recipe library. The builder writes only its working-copy output and removes only obsolete files recorded in its own previous generated manifest. It does not modify the original Studio source or deploy a site. Existing recipe and wrapper pages are preserved.

## Rebuild
1. Prepare the canonical existing-resource records and readable wrappers with `python3 work/prepare_resources.py` when that preparatory script or source data changes.
2. Review `work/resources.json`, `work/voice-tools.json`, `work/image-tools.json` and `work/video-tools.json`. Tool ratings carry their task and domain before merging.
3. Run `python3 work/build_studio.py` from the package directory. The output is `outputs/studio/`. CSS and browser behavior come from `work/studio-ui.css` and `work/studio-ui.js`.
4. Validate generated local links, JSON, SVG XML, Python and JavaScript syntax. Check browsing and filters at desktop and phone widths before publishing.
5. Use `outputs/studio/generated-files.json` for builder-owned files and `work/wrapper-files.json` for preparatory wrappers. Deployment is a separate operation.

## Shared exports
`catalogue/catalogue.json` contains canonical resource and tool records. `catalogue/catalogue.md` provides a short text index for AI context. `catalogue/method.html` explains ratings, evidence and how to use the catalogue with an AI.