Ai

Scrapling v0.4.7 Adds Visual Browser Context for AI Agents

Scrapling v0.4.7 Adds Visual Browser Context for AI Agents

Scrapling v0.4.7 is a focused release aimed at making AI-driven browser workflows more practical and observable. The standout addition is a new screenshot MCP tool that lets models capture and inspect live page images during an active browser session. Alongside that, the release improves session management with custom session IDs and resolves a pair of bugs affecting session stability and AI-targeted CLI usage.

What Changed

The biggest update in v0.4.7 is the new screenshot MCP tool. It captures a page and returns it as a real MCP ImageContent block, allowing the model to visually inspect the rendered page instead of relying only on text or DOM state. The feature works with an existing browser session, meaning users must first call open_session in either dynamic or stealthy mode and then pass the corresponding session_id. It supports PNG and JPEG output, full-page screenshots, JPEG quality settings, and readiness controls such as wait, wait_selector, network_idle, and timeout.

Another important workflow improvement is support for a custom session_id parameter in open_session. Instead of defaulting to a random 12-character hex identifier, teams can now assign human-readable names such as search or checkout. This should make multi-step automation and AI agent orchestration easier to manage, especially when several browser sessions are active at once.

On the reliability side, v0.4.7 fixes FetcherSession state corruption as well as a lazy session close crash, both of which address stability issues in session handling. The release also fixes a TypeError triggered when using the CLI's --ai-targeted flag with HTTP commands, specifically around an unexpected block_ads argument being passed into Session.request().

The release also includes a new Brazilian Portuguese README translation, extending project accessibility for a broader developer audience.

Why It Matters

The new screenshot MCP capability is the clearest signal of where Scrapling is heading: more direct support for AI agents that need visual awareness during browser automation. For teams building agentic workflows, this matters because rendered screenshots can help models validate layouts, spot UI changes, and make better decisions in real browser contexts.

Custom session naming is a smaller feature on paper, but it can meaningfully improve session traceability in production automation. Naming sessions by task or funnel stage makes debugging and orchestration more intuitive, particularly in complex scraping or browser interaction pipelines.

The bug fixes are also significant for real-world use. Stability issues in session state and compatibility problems with AI-targeted CLI flags can undermine reliability in automated workflows. By addressing both, Scrapling v0.4.7 strengthens the foundation for developers using the project in AI-assisted browsing, scraping, and data extraction scenarios.

Official Source: https://github.com/D4Vinci/Scrapling/releases/tag/v0.4.7