Skit SDK

Factory functions

import { defineScheduledScraper, defineDaemonScraper } from "@minato/skit";

Export the result as your module's default export.

Function	Use when
`defineScheduledScraper`	`run()` does one pass and exits
`defineDaemonScraper`	`run()` maintains a persistent connection

Pass TConfig as a type parameter to type your config:

type Config = { maxPages: number; delayMs: number };
export default defineScheduledScraper<Config>({ ... });

Context

run() receives a single context object:

Property	Type	Description
`ingest`	`IngestClient`	Buffer and flush torrents
`config`	`TConfig`	Merged config (dashboard overrides applied)
`signal`	`AbortSignal`	Aborted on pause or stop
`status`	`StatusReporter`	Report progress to the dashboard
`meta`	`{ id, name, version }`	Scraper identity
`flaresolverr`	`FlareSolverr`	Pre-configured Cloudflare bypass client

`ingest`

ingest.add(torrent: TorrentInput): void

Buffers a torrent. Minato flushes in batches automatically; you never call flush() yourself. infoHash is the global deduplication key: the same hash is never stored twice, regardless of which scraper submitted it.

`TorrentInput`

Field	Required	Type	Notes
`infoHash`	✅	`string`	40-char hex info hash
`title`	✅	`string`	Release name
`size`	✅	`number`	Total size in bytes
`source.name`	✅	`string`	Label shown in the UI. Use `meta.name`.
`seeders`		`number`
`leechers`		`number`
`magnet`		`string`	Full magnet URI
`category`		`string`	Tracker-specific category label
`publishedAt`		`string`	ISO 8601
`files`		`Array<{ filename, size }>`
`source.url`		`string`	Torrent detail page
`source.origin`		`string`	Tracker or index name
`source.originUrl`		`string`	Tracker homepage

`config`

The result of merging minato.defaultConfig with any overrides the admin set in the dashboard. Dashboard values always win; you can't override them from code. Treat config as read-only at runtime.

const { maxPages, delayMs } = config;

`signal`

Fires on both pause and stop. The scraper can't distinguish between them; that's Minato's concern. Check at loop boundaries and return cleanly when aborted.

for (const page of pages) {
  if (signal.aborted) break;
  // ...
}

Buffered torrents flush automatically before exit. A paused scraper can be resumed on demand; a stopped one waits for its next scheduled run.

`status`

status.update({
  phase?: "idle" | "running" | "paused" | "error",
  progress?: { current: number; total?: number },
  message?: string,
})

All fields optional. Failed calls are silently ignored.

`meta`

meta.id       // "my-scraper"
meta.name     // "My Scraper"
meta.version  // "1.0.0"

Use meta.name as source.name in ingest.add() so the source label in the UI stays consistent with the dashboard name.

`flaresolverr`

A pre-configured FlareSolverr client for sites protected by Cloudflare. Use ffetch as a drop-in replacement for fetch, routing the request through FlareSolverr and returning a standard Response.

import { ffetch } from "@minato/skit";

export default defineScheduledScraper({
  async run(ctx) {
    const res = await ffetch(ctx, "https://protected-site.example/list");
    const html = await res.text();
    // parse and ingest...
  },
});

ffetch supports GET and POST only. It throws if FlareSolverr is unreachable or returns an error, so the failure surfaces as a normal scraper crash rather than silently producing empty results.

Skit SDK

On this page