Changelog.

Search now feels more personal and more grounded. You can guide results using your current setup and carry that context through the rest of the flow.

• Context-aware search: You can now refine search using agent, stack, and environment preferences directly in the main experience.
• Smarter continuity: Your context now follows you from search into skill pages, compare views, and install flows.
• Clearer activation checks: Install flows now show lightweight preflight guidance so it is easier to spot what looks ready and what needs a closer look.

We've completely overhauled our AI synthesis layer to be faster, smarter, and more transparent when evaluating skills.

• Real-time Analysis: Skill synthesis now streams word-by-word instantly, entirely eliminating visual wait times.
• Trust Percentiles: Added clear, human-readable risk explanations and comparative trust percentiles to help you evaluate safety before ever activating a skill.
• Adaptive Intent: The search engine now detects query complexity on the fly to provide deeper, more nuanced recommendations for complex tasks.

We completely rebuilt the core search pipeline to drastically improve how MaTE Skills understands natural language.

• Advanced Semantic Quality: Upgraded our core embedding models to deliver unparalleled accuracy when searching by concept rather than exact keyword.
• Dynamic Retrieval: The engine now intelligently balances exact keyword matching against deep contextual understanding.
• Lightning Fast Architecture: Implemented deep optimization and caching layers ensuring immediate search results no matter the load.

Resolved edge-case login friction and hardened the platform's infrastructure.

• Resolved bugs that prevented some users from authenticating correctly.
• Improved system resistance to high-traffic spikes and aggressive API crawling.
• Smoothed the entire redirection flow for a seamless sign-in experience.