Lexora Buddy: starší verze ke stažení

This release continues to refine the Agent experience. Mention companions by clicking their avatars or names, and view local workspace images directly in conversations. Quoted messages retain the original companion’s name, while sidebar indicators distinguish active execution, pending answers, and approval requests.
We’ve addressed long-conversation scrolling, reading-position restoration, duplicate queued messages, slow browser navigation, and compatibility with certain image-generation services. The independently updated Local Models plugin also improves local imports with manual capability selection, optimized GGUF reading, and Qwen3.5 model-structure validation.

The Local Models plugin is here—from discovery and downloads to running models on your own computer. This is the headline feature of 0.20.0: search Hugging Face, import existing model files, and bring local chat, image understanding, speech transcription, read-aloud, image generation, and OCR into your everyday tasks.
We’ve also refined task management and daily interactions. Task history now supports pagination and global search, companions can have their own reasoning-effort settings, and installed plugins have a more direct update entry point. This release also fixes issues with conversation reading, text pasting, system speech, and multi-companion collaboration.

This is a patch release for 0.19.0. The previous release introduced the ability for companions to ask you questions, but the question cards’ styling and interactions weren’t where we wanted them to be. We’re following up with a refined layout, clearer visual hierarchy, and one-at-a-time question navigation to make answering more focused and comfortable.
We also addressed performance issues to make long conversations and split-pane resizing smoother, alongside fixes for delayed supplementary messages and loading and reading-position issues after editing or regenerating replies.
There’s still room for improvement, and we’ll keep refining these experiences in future releases.

This release focuses on making everyday use smoother, clearer, and less disruptive. From starting the app and sending messages to reading replies aloud, generating images, and switching views, these refinements reduce unnecessary waiting and repeated checks so you can stay focused on your task.
Agents can now ask you questions within a task when they need information or a decision, collecting your answers through choices or text. Browser operations also have separate execution pages that open in the background by default, reducing interruptions to what you are viewing.
You can also choose between Stable and Beta update channels and adjust tool invocation to suit your model, adding flexibility to your familiar workflow.

You can now capture ideas with voice input and listen to companion replies. Image understanding, generation, and OCR also have centralized settings, so you no longer need to keep switching conversation models to work with images.
Explore community plugins in the Plugin Marketplace to add features you enjoy. This release introduces a DSH-inspired task status bar, alongside refinements to downloads, concurrent execution, the composer, and file operations.

This release makes full access live up to its name. Tasks are no longer confined to workspace directories or saved directory grants, and commands run directly on the host without sandbox network isolation. Available tool operations—including file and credential access, commands, network requests, and MCP calls—are automatically approved, aligning the mode's actual capabilities with its meaning.
We also refined everyday interactions and polished a few details to make Lexora easier to use.

This update improves handoffs and continuation in multi-agent workflows. Companions can build on previous context and results to refine or rework an assignment, share relevant materials, and communicate directly with one another, reducing repeated explanations. Execution details now distinguish task requirements, delivery evidence, and review outcomes.
It also fixes issues with companion identity, coordination waits, and task completion feedback, while improving @mention routing and independent scheduling, access to previous reports, result actions, and usage display.

This release focuses on Lexora Office and document workflows. Alongside the Office plugin, we added file-opening preferences, content references, bundled skills, clipboard access, and private document storage—connecting document editing, conversations, and AI-generated deliverables so companions can work with documents rather than only discuss them.
This version also introduces a Windows ZIP distribution, customizable companion stickers and notification sounds, plus improvements to reading position, companion collaboration, and the built-in browser.