# Resensa — AI for YouTube videos

> Resensa is a suite of tools for working with YouTube video transcripts:
> a browser extension that embeds AI summaries / Q&A right inside
> youtube.com, a standalone search app, and an MCP server so AI
> assistants (Claude, Cursor, ChatGPT, Codex, Windsurf) can query the
> same backend programmatically.

## Surfaces

- **Landing** (this site) — <https://resensa.io/> — product overview,
  pricing, install links, privacy/terms.
- **Browser extension** — <https://chromewebstore.google.com/detail/resensa-%E2%80%94-ask-youtube-ai/cggjdmmehmbjkndclkfflgencnfkjaol>.
  Adds summaries, Q&A with quoted timestamps, and a structured table of
  contents to every YouTube watch page.
- **Search app** — <https://resensa.io/search> — subscribe to YouTube
  channels / playlists / videos and full-text-search across their
  transcripts. Per-browser anonymous library, no signup.
- **MCP server** — <https://search-mcp.resensa.io/mcp> (Streamable HTTP).
  Exposes the search backend as Model Context Protocol tools. Agents
  should call this directly instead of scraping HTML.

## For AI agents

The search app is a SPA — its HTML contains no static content worth
crawling. To programmatically query the index, connect to the MCP
server above. Available tools include `search`, `add_sources`,
`get_chunks_by_range`, `sync_source`, `resolve_urls`.

### Quick MCP config (Cursor / Claude Desktop / Windsurf)

```json
{
  "mcpServers": {
    "resensa": {
      "type": "http",
      "url": "https://search-mcp.resensa.io/mcp"
    }
  }
}
```

### Codex config (~/.codex/config.toml)

```toml
[mcp_servers.resensa]
url = "https://search-mcp.resensa.io/mcp"
```

See <https://resensa.io/search/llms.txt> for the search app's own
agent-readable description and <https://resensa.io/AGENTS.md> for the
full agent guide.

## Contact

support@resensa.io