# Demo Platform (working name)

Self-hosted, open-source interactive demo platform. One capture → interactive
clickable demo, step-by-step guide, narrated video. MIT licensed, built and
owned by DiligenceWorks Pte. Ltd.

**Status: v0.1 — early alpha.** Interactive demos (image steps) work end to end:
capture with the browser extension → refine in the editor → publish → share/embed.

## Why

The interactive-demo SaaS category (Arcade, Storylane, Navattic, Supademo) is
closed-source and $500–1,200/mo, and your demos — unreleased UI, prospect
analytics, lead data — live in someone else's cloud. This runs on your
infrastructure. Nothing leaves it.

## AI without a bundled model

The platform **never calls an LLM**. It ships an **MCP server** (`POST /mcp`,
Streamable HTTP): attach your own agent — Claude, an IDE, your own runtime —
with a scoped API key, and it can list demos, write annotations, place
hotspots, reorder steps, read analytics, and (only with an explicitly granted
`publish` scope) publish. Draft-only by default; the human decides who may
publish. Viewer-supplied data (leads) is wrapped in untrusted-data markers.

## Quick start

```bash
docker compose up -d --build
docker compose logs app | grep -A2 "FIRST RUN"   # your admin API key, shown once
```

Open `/editor`, paste the key. Create a demo, upload screenshots as steps (or
install `extension/` as an unpacked Chrome extension and record a real
click-through — hotspots are pre-placed from your actual clicks), drag hotspots,
write tooltips, publish. Share `/p/<token>`, embed with an `<iframe>`. Optional
viewer identification: append `?v=<viewer-token>` to the link you send.

## Privacy defaults

No IP addresses stored. No fingerprinting. Viewer identity only via explicit
share-link tokens. Lead deletion endpoint included.

## API scopes

`read` · `author` · `publish` · `admin` — keys default to `read,author`.

## Roadmap

DOM-fidelity capture (rrweb serialization, asset pipeline), step-guide export,
narrated video render, desktop capture agent. See DESIGN docs in the DW KB.

## License

MIT © 2026 DiligenceWorks Pte. Ltd.