tauri (2.9.5 → 2.10.2)
diff --git a/src/pages/docs.astro b/src/pages/docs.astro
index 825b0c5..22e2196 100644
--- a/src/pages/docs.astro
+++ b/src/pages/docs.astro
@@ -103,10 +103,10 @@ import BaseLayout from '../layouts/BaseLayout.astro';
@ for special commands@ for special commandsThe AI automatically knows about the file you are currently looking at. You don't need to copy-paste code into the chat.
On X11-based desktops (e.g., Openbox, i3, Fluxbox), window capture has the following limitations:
picom or compton, these limitations are largely eliminated — compositors maintain off-screen buffers for all windows, enabling capture of obscured or unfocused windows.picom or compton, these limitations are largely eliminated - compositors maintain off-screen buffers for all windows, enabling capture of obscured or unfocused windows.These limitations do not apply to Wayland desktops (GNOME, KDE Plasma 6) or macOS, which use compositing by default.
diff --git a/src/pages/index.astro b/src/pages/index.astro index c7ef381..4e8de65 100644 --- a/src/pages/index.astro +++ b/src/pages/index.astro @@ -30,7 +30,7 @@ const version = pkg.version ?? "";- What you see is just the interface. The real magic lives in the Zaguán Coder Daemon—a Go-powered backend where all AI reasoning happens. + What you see is just the interface. The real magic lives in the Zaguán Coder Daemon - a Go-powered backend where all AI reasoning happens. Its standout feature is Unlimited Context: instead of hitting token limits, we intelligently compress your conversation history and tool results into retrievable artifacts. The AI always gets exactly what it needs to work efficiently, without the bloat.
The actual brain of the operation. A Go server that handles all AI logic, system prompts, and context management. This is where the heavy lifting happens—your editor stays lightweight and responsive.
+The actual brain of the operation. A Go server that handles all AI logic, system prompts, and context management. This is where the heavy lifting happens - your editor stays lightweight and responsive.
We've solved the token limit problem. Chat history and tool results are actively compressed into artifacts. The AI retrieves only what it needs—efficient, focused, and without arbitrary limits.
+We've solved the token limit problem. Chat history and tool results are actively compressed into artifacts. The AI retrieves only what it needs - efficient, focused, and without arbitrary limits.
The GUI contains zero AI logic—it's purely interface. All intelligence lives in the Coder Daemon. This means a lighter client and a more powerful, updatable backend.
+The GUI contains zero AI logic - it's purely interface. All intelligence lives in the Coder Daemon. This means a lighter client and a more powerful, updatable backend.
Long conversations are automatically distilled into retrievable artifacts. The AI knows when and how to access this context—maintaining coherence without the overhead.
+Long conversations are automatically distilled into retrievable artifacts. The AI knows when and how to access this context - maintaining coherence without the overhead.
Multi-patch operations are atomic — all patches are validated before any are applied. If any patch fails, no changes are made.
+Multi-patch operations are atomic - all patches are validated before any are applied. If any patch fails, no changes are made.
The 50KB / 2000 line limits apply only to tool results — what zblade sends back to the model after executing a tool (e.g., the output of read_file). These limits only apply in local mode, when the Zaguán Coder Daemon isn't handling storage.
The 50KB / 2000 line limits apply only to tool results - what zblade sends back to the model after executing a tool (e.g., the output of read_file). These limits only apply in local mode, when the Zaguán Coder Daemon isn't handling storage.
There are no size limits on incoming tool calls from the model. zblade will accept a write_file with any size content. The write_file handler itself has no caps — it writes whatever content string it receives directly to disk.
There are no size limits on incoming tool calls from the model. zblade will accept a write_file with any size content. The write_file handler itself has no caps - it writes whatever content string it receives directly to disk.