Built-in research source integrations for @debriefer/core: Wikipedia, Wikidata, web search, news, archives, and more.
npm install @debriefer/core @debriefer/sources| Category | Count | Cost | Sources |
|---|---|---|---|
| Structured data | 2 | Free | Wikipedia, Wikidata |
| News | 22 | Mostly free | AP, BBC, Reuters, NPR, Guardian, NYT, Washington Post, LA Times, Rolling Stone, Britannica, TCM, and more |
| Web search | 4 | Mixed | DuckDuckGo (free), Google, Bing, Brave (API key required) |
| Books | 2 | Mixed | Open Library (free), Google Books (API key required) |
| Archives | 4 | Free | Chronicling America, Trove, Europeana, Internet Archive |
| Obituary | 2 | Free | Find a Grave, Legacy.com |
News sources marked "mostly free" use DuckDuckGo site-search under the hood and require no API keys. The Guardian and New York Times sources use their official APIs and require keys.
import { ResearchOrchestrator, NoopSynthesizer } from "@debriefer/core"
import { wikipedia, apNews, guardian, wikidata } from "@debriefer/sources"
const orchestrator = new ResearchOrchestrator(
[
{ phase: 1, name: "Structured", sources: [wikidata(), wikipedia()] },
{ phase: 2, name: "News", sources: [apNews(), guardian()] },
],
new NoopSynthesizer()
)
const result = await orchestrator.debrief({ id: "ada-lovelace", name: "Ada Lovelace" })@debriefer/sources exports utilities useful when building custom sources:
| Utility | Description |
|---|---|
fetchPage |
HTTP fetch with timeout, User-Agent, and redirect handling |
extractArticleContent |
Mozilla Readability extraction from raw HTML |
sanitizeSourceText |
Clean extracted text for use as research findings |
htmlToText |
Convert HTML to plain text |
searchDuckDuckGo |
Run a DuckDuckGo search and return result URLs |
See the monorepo README for full documentation including orchestrator configuration, reliability scoring, and adding custom sources.
MIT