Understanding Deliverables
What you get
Scriptonia produces a project blueprint: structured docs, specs, LLM prompts, and infra notes. It is not a fully runnable app out of the box. You use the prompts and specs to generate or implement code, then deploy using the infra guidance.Common artifacts (all platforms)
| Artifact | Purpose |
|---|---|
| README.md | Auto-generated project overview: session, platform, description, features, target audience, tech requirements, execution summary, and pointer to FOLDER_STRUCTURE.md. |
| FOLDER_STRUCTURE.md | Tree of the project (folders and files). |
| REVIEW_REPORT.md | ReviewAgent output: structure check, completeness, and any issues. |
By platform
Website
docs/— PRD.md, PLAN.md, SEO_CONTENT.mdfrontend/— nextjs.md, pages.md, components.mdfrontend/prompts/— generate_static_site.txtcontent/— blog/, assets/infra/— hosting.md, cdn_caching.mdmeta/— github_push.yaml
Web Application
docs/— PRD, PLAN, HUMAN_REVIEW, PROJECT_RULESfrontend/— react_nextjs, components, state_managementfrontend/prompts/— generate_frontend.txtbackend/— node_express, services, authbackend/apis/— openapi.mddb/— schema, migrations, orm_modelsinfra/— terraform, ci_cd, monitoring, secrets_managementtesting/— unit_tests, api_testsmeta/— github_push.yaml
Android
android/— kotlin_ui, fragments, navigationandroid/prompts/— generate_android.txtbackend/— api_contracts, push_notifications, node_firebasedb/— firebase_schemainfra/— gcp_deploymeta/
iOS
ios/— swiftui, storyboards, navigation, permissionsios/prompts/— generate_ios.txtbackend/— api_contracts, push_notifications, node_firebasedb/— local_storage (CoreData)infra/— ci_cdmeta/
File types
- Markdown (.md) — PRD, PLAN, specs, infra, README, FOLDER_STRUCTURE, REVIEW_REPORT
- YAML (.yaml) — github_push, CI/CD or config
- Prompts (.txt) — LLM prompts used to generate actual code
Downloading Projects
When the download is available
On the Execution page, after:- Review passed —
review_passed === true - ZIP generation finished —
POST /api/projects/generatecompleted - “Download Project ZIP” is shown and “ZIP file ready!” appears
How to download
- Click “Download Project ZIP” on the Execution page
- App requests
GET /api/projects/download?sessionId=... - Server returns a ZIP:
- From ZipFile table, or
- Built from project directory
- File is saved as
<projectName>.zipor<sessionId>.zip
What’s in the ZIP
- Same layout as in Understanding Deliverables
- All paths under a single project root
- No extra
sessionIdfolder unless defined by the template
Completion page
The Completion page (/completion?sessionId=...) shows a summary and next steps.
- Main delivery (download, GitHub) happens on Execution
- Completion is a follow-up view
- You can start a new project from there
If download fails
- “Project not found” / “Please generate files first” — Execution did not finish or Review failed
- “Failed to download ZIP file” — Network or server error; retry or contact support with
sessionId
GitHub Integration
Connect GitHub
- On the Execution page, click “Connect to GitHub”
- Redirect to
/api/github/authorize?sessionId=... - GitHub OAuth (scope: repo)
- Redirect to
/api/github/callback, which:- Exchanges code for token
- Sets cookies
- Redirects back to Execution
- Execution page shows Connected as <username> and Upload to GitHub
Upload to GitHub
- Click “Upload to GitHub”
- Set:
- Repository name
- Description (optional)
- Private / public
- Confirm →
POST /api/github/upload - Server:
- Creates repo
- Reads project files
- Commits and pushes
- Repo opens in a new tab
Requirements
- GitHub OAuth configured
- Project files exist after generate
github_access_tokencookie present
If upload fails
- GitHub not connected — Complete OAuth
- Project files not found — Generate did not complete
- Failed to create repository — Name conflict or missing scope
Using Generated Code
Blueprint, not a full app
The ZIP contains specifications, prompts, and infra, not runnable code.Suggested order
- README.md
- docs/PRD.md, docs/PLAN.md
- Specs (frontend, backend, db)
- Prompts
- infra/
Prompts
- Located in
frontend/prompts/,ios/prompts/,android/prompts/ - Each
.txtis an LLM prompt - Use with matching specs and PRD/PLAN
- Adjust as needed
Environment and secrets
- Use
.env.exampleas a template - Fill real values in
.envor deployment config - Do not commit secrets
Deploy
- Follow infra/ docs
- Use meta/github_push.yaml for CI/CD
- Website: Vercel/Netlify
- Web apps: Docker, AWS, Vercel
- Mobile: App Store Connect, Play Console, Xcode Cloud, GCP
Next Steps After Delivery
- Review
- Customize
- Generate runnable code
- Configure
- Deploy
- Iterate
- New project