# Launch Plan

## Repository Positioning

Name:

`privacy-redactor`

One-line description:

> Local-first sensitive information redaction before sharing text with AI tools, apps, or people.

GitHub topics:

- `privacy`
- `pii`
- `redaction`
- `local-first`
- `web-component`
- `ai-safety`
- `data-loss-prevention`
- `secrets`
- `logs`
- `browser`

## README Hook

Lead with a concrete before/after example:

```text
Email: yang@example.com
api_key=sk_test_1234567890abcdefghijklmnop
Booking reference: ABC123

Email: [EMAIL]
api_key=[API_KEY]
Booking reference: [BOOKING_REFERENCE]
```

## Public Demo

Publish `demo/` with GitHub Pages after the repository is renamed or created.

The demo should state:

- Runs in your browser.
- No upload.
- No network request.
- Synthetic examples only.
- Industry examples use fake banking, patient, legal, HR, and student identifiers.

## Early Audience

- Hacker News "Show HN".
- GitHub / npm privacy tooling communities.
- AI builder communities.
- Browser extension builders.
- Third-party web apps that need a no-backend local redaction widget.
- Developer tooling communities.
- Chinese-speaking AI and privacy communities.

## Good First Issues

- Add deeper healthcare mode fixtures.
- Add deeper finance mode fixtures.
- Add deeper legal mode fixtures.
- Add locale-specific ID patterns beyond the current China resident identity card validator.
- Add PDF.js extraction adapter.
- Add Tesseract.js screenshot OCR adapter.
- Harden browser extension popup UX and add cross-browser notes.
- Add runnable Svelte and Next.js examples.
- Add Web Component entity filtering and confidence explanations.

## Credibility Moves

- Keep the core dependency-free.
- Publish limitations clearly.
- Use synthetic fixtures only.
- Run `npm run evaluate` before publishing rule changes.
- Keep local-only behavior as default.
- Use false-negative and false-positive issue templates.
- Document safe integration boundaries.
