AI BriefingSavage MCP Released: Limited Browser Control Tools for LLMs
Key point
Open-source tools Savage MCP and Savage Scraper have been released to minimize browser control permissions for LLM agents.
Details
Developer Dominik Duda has released two open-source tools with strictly limited permissions to reduce security risks associated with LLM agents.
Savage MCP and Savage Scraper
- Savage MCP: A locally running MCP server that allows LLMs to read only permitted web pages via an existing Chrome session.
- Savage Scraper: A Chrome extension that converts rendered web pages into simplified, LLM-friendly HTML.
Security-Centric Design
The extensive control features provided by existing browser automation APIs are intentionally excluded. Agents are limited to opening permitted URLs, scraping dedicated tabs, and reporting status; arbitrary clicks, typing, form submissions, and JavaScript execution are blocked. Website access permissions are configured locally at the host and path level, and the Chrome extension independently verifies them. Both projects are released under the MIT license.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Foremerge v0.5.0 Released: Rust-Based Open-Source Tool Detects Intent Conflicts Between Parallel AI Coding AgentsJevper Released: A Jev Classification Wrapper for OpenAI-Compatible Models with Automatic Logprobs Detection