Website Context Extractor is a tool that automatically finds, gathers, and attaches context to your translation strings from your live website. It uses AI to detect where each string appears and provides accurate, real-time context for every translation.

Key Features

• Use an AI model and a customizable prompt to analyze web pages and automatically add textual context to strings.
• The crawler scans your specified website or URL, collecting source strings and mapping them to their actual on-page usage.
• Run the app on demand in Manual Mode or automate extraction with Webhook Mode.
• Filter strings by file, label, or date. Configure Request Modifications to add URL GET parameters, enabling the app to crawl protected pages.
• In Webhook Mode, the app can automatically trigger pre-translation after context is added, improving your workflow.
• Maintain quality by reviewing, editing, or removing extracted contexts before saving them to Crowdin.

How It Works

The extractor has two modes: manual and webhook.

1. Manual Mode

Use this mode for testing, targeted extraction, or processing specific file updates.

• Input: Provide the Website URL (a single page or the full site) and specify your AI Model (gpt-4.1 for example) and Prompt to guide the context analysis.
• Filter Strings: Apply filters by file, label, or date range to target the exact strings that need context.
• Advanced Access: Configure optional Request Modifications (e.g., adding GET parameters) to crawl protected pages.
• Review: After processing, you can review, edit, or delete the suggested contexts before saving them to your Crowdin project.

2. Webhook Mode

This mode runs automatically as a background process, ensuring new strings always have context.

• Trigger: The agent is automatically triggered whenever a new file is uploaded or an existing file is updated in Crowdin.
• Auto-Context: It extracts context for all context-less strings.
• Workflow Acceleration: You can optionally enable auto-pre-translation to run immediately after the context is successfully added, moving new content straight to the review stage.

Screenshots