decant-core

A shared extraction layer for the modern web — including AI conversations.

Every AI chat exporter ends up solving the same problem: extracting conversations from ChatGPT, Claude, Gemini, Perplexity, DeepSeek and other constantly changing AI interfaces.

And every time one of those platforms changes its UI, seriously re-renders a message, or ships a new feature, somebody's parser breaks.

decant-core provides reusable parsers, platform detection, and web article extraction so developers don't have to build and maintain the same fragile parsing layer over and over again.

Origin & Purpose

Existing chat exporters often suffer from two major flaws: they break whenever platform DOMs update, and many route user conversations through third-party servers.

decant-core was created to solve both at the foundation. Originally built to power local-first extensions like AI Chat Exporter and Decant, it decouples fragile platform parsing from presentation. By sharing this engine under AGPL-3.0, any browser extension, web clipper, archiver, or research tool can rely on a maintained, local-first extraction layer instead of reverse-engineering AI platforms in isolation.

npm install decant-core

Use it as the parsing layer underneath your own:

• chat exporters
• browser extensions
• web clippers
• research & data-extraction tools
• content archivers
• knowledge-management and PKM applications

Fix platform parsing once, and let the ecosystem benefit from the fix.

Why decant-core?

AI platforms don't expose stable public APIs for reading conversation history. No matter what you build, to extract a ChatGPT thread you need to walk the DOM, read internal RPC payloads, or traverse React component trees — and redo it when the frontend changes.

Maintaining that per-platform logic in every exporter is wasteful and fragile. decant-core centralizes it:

• ✅ 19 AI chat platform parsers with normalized output — you get structured messages, models, metadata and Markdown, not DOM soup.
• ✅ Web article extraction — Mozilla Readability, Defuddle, and Article-Extractor run in parallel and arbitrate by content-quality scoring.
• ✅ Detection utilities — tell an "AI chat page" apart from a "regular web page" before you decide which parser to run.
• ✅ Math & Markdown handling — LaTeX normalization plus GFM tables/code fencing that survive round-trips into Obsidian, Logseq and Notion.

The payoff is maintenance: when a platform changes, the fix happens once, in one place, instead of being independently reimplemented across dozens of projects.

Quickstart

1. Extract an AI conversation

import { detectPlatform, isAiChatUrl } from "decant-core";

if (isAiChatUrl(window.location.href)) {
  const match = detectPlatform(window.location.href);

  if (match?.parser && match.parser.isAvailable(window.location.href)) {
    const result = await match.parser.parse();
    // result.title
    // result.messages  -> [{ role: 'User' | 'Assistant', content, ... }]
    // result.model
    // result.metadata  -> platform-specific extras
  }
}

2. Extract a regular web article

import { extractArticle } from "decant-core";

const article = await extractArticle(document /* or an HTML string */, {
  url: window.location.href,
});

// article.title, article.author, article.published
// article.markdown       -> clean, ready-to-use Markdown
// article.content        -> the body without the title prefix
// article.engine         -> 'readability' | 'defuddle' | 'raw'

3. Detection

import { detectPlatform, isAiChatUrl, AI_CHAT_DOMAINS } from "decant-core";

isAiChatUrl("https://chatgpt.com/c/abc-123"); // -> true
const detected = detectPlatform(url); // -> { type: 'ai-chat', platform: 'ChatGPT', parser }

Subpath imports

// Individual parsers (tree-shake the rest)
import { ChatGPTParser } from "decant-core/ai/chatgpt";
import { ClaudeParser } from "decant-core/ai/claude";
import { GeminiParser } from "decant-core/ai/gemini";

// Web & article extraction
import { extractArticle, ArticleParser, scoreContent } from "decant-core";

// Detection
import { detectPlatform, isAiChatUrl, parsers } from "decant-core";

// Utilities
import { convertToMarkdown, cleanMarkdown } from "decant-core";
import { normalizeLatexMath } from "decant-core";

Supported Platforms

19 AI chat platform parsers plus generic web article extraction:

| Platform | Parser | Extraction strategy | | :---------------------------------- | :------------------------ | :----------------------------------------- | | ChatGPT | ChatGPTParser | DOM + internal API | | Claude | ClaudeParser | DOM + internal API + React fiber | | Google Gemini | GeminiParser | DOM + batchexecute RPC | | Microsoft Copilot | CopilotParser | DOM | | Perplexity | PerplexityParser | Internal API + DOM | | DeepSeek | DeepSeekParser | DOM + internal API | | Qwen | QwenParser | DOM | | Meta AI | MetaParser | Internal API + DOM | | Mistral / Le Chat | MistralParser | DOM | | Proton Lumo | LumoParser | DOM | | Z.ai | ZAiParser | Internal API + DOM | | Grok | GrokParser | Internal API + DOM | | Google AI Studio | GoogleAIStudioParser | DOM | | NotebookLM | NotebookLMParser | DOM | | Google Search AI (AI Overviews) | GoogleSearchAIParser | DOM | | Gemini Cloud Assist | GeminiCloudAssistParser | DOM | | Joyland | JoylandParser | DOM | | Chub | ChubParser | DOM | | Duck.ai (DuckDuckGo AI) | DuckAIParser | DOM | | Generic Web Article | ArticleParser | Readability + Defuddle + Article-Extractor |

All parsers extend the base ChatParser interface — a consistent isAvailable(url) + normalized parse() contract. For the full extraction-strategy breakdown and maintenance model, see SUPPORTED_PLATFORMS.md, also published as the platform matrix on the developer docs site.

License

decant-core is licensed under the GNU Affero General Public License v3.0 (AGPL-3.0-only) with a FOSS Linking Exception, alongside a Commercial License option.

FOSS Linking Exception (Open Source)

Permission is granted to link, import, or bundle decant-core into projects distributed under any OSI-approved open source license (including MPL-2.0, MIT, Apache-2.0, and BSD) and distribute the resulting work under that project's license, without requiring the enclosing project to be licensed under AGPLv3. Any modifications directly made to decant-core source files remain subject to AGPLv3.

Commercial License

If you wish to use decant-core in closed-source, proprietary, or commercial software that cannot comply with the AGPLv3, a commercial license is available. Please contact [email protected] for licensing terms.

Third-Party Test Fixtures Notice

Sample test fixtures located in tests/fixtures/ consist of third-party DOM snapshots and API response excerpts retained solely for automated regression testing and platform interoperability under fair use principles. They are excluded from the project's AGPL-3.0 license. See tests/fixtures/README.md for details.

Used by

• AI Chat Exporter — export, archive and transfer AI conversations between platforms.
• Decant — the distraction-free web clipper and research batcher.

These products are demonstrations of the library, not its purpose. Yours can be next — see CONTRIBUTING.md.

Development

Building, testing, and extending the library is covered in DEVELOPMENT.md; platform contributions follow the parser pattern and CLA in CONTRIBUTING.md.