Model benchmarks, architectural deep dives, and honest field notes from building an evidence-grounded AI editor.
Zaguán Blade v1.0.0 arrives on August 1, 2026. This is the story of five discarded starts, the Rust and Tauri combination that finally clicks, zcoderd becoming the system's core, and the releases that turn a personal daily-driver goal into a real product.
A 12-minute session, 17 tool calls, one surgical CSS patch. We pulled the logs from a real Zaguán Blade session to show what happens when a model uses the Symbols Index the way it was designed — and how zcoderd quietly kept it on rails the whole time.
A few-day war story indexing all 469, 000 files of Firefox on a USB spinner, the crashes, the hangs, the confidently wrong guesses, and how live profiling (and slow storage) turned a 2.5-hour ordeal into 30 minutes.
Zaguán Blade's Symbols Index helps AI coding agents understand code structure before they read everything, giving models the right context at the right time without turning every task into a token furnace.
AI coding agents can get stuck inside stale assumptions. Zaguán Blade's Cognitive Interrupt pauses the loop, forces a reframe, and helps the model inspect evidence before trying again.
A quick look at two new Zaguán Blade features: controlling Blade from Telegram and capturing, annotating, and sharing screenshots directly from the chat composer.
A technical introduction to Zaguán Blade: the open-source, Rust-first AI code editor built around agentic workflows, disk-first changes, backend-authoritative state, and a protocol-driven architecture instead of another VSCode fork.
A practical look at the Zaguán Blade Skills catalog: why we chose this particular mix of workflows, why skills are more than prompts, and why Vibe Coding needs reusable judgment.
Most AI coding assistants answer the questions you ask. Zaguán Blade's Blindspot Radar helps surface the questions you were not asking yet.
The current approach to LLM-based security auditing is fundamentally flawed — it produces noise instead of signal. We built something different, and we're publishing the full methodology.
A semi-technical introduction to the Zaguán Language Protocol: why we built it, where it fits inside Blade, and why AI coding needs a different kind of protocol than traditional IDE tooling.
A light, semi-technical introduction to zcoderd, the server-side brain behind Zaguán Blade, and why splitting intelligence from execution makes AI coding feel more useful.
Exploring Mistral Devstral 2 - a powerful new AI model that brings enhanced coding capabilities to Zaguán Blade.
A data-driven look at how context compression and selective retrieval reduce real model costs in long-running coding sessions.
Learn how to leverage Local AI models through Ollama and create effective system prompts that work well within the constraints of limited context windows and hardware limitations.
Announcing our new blog where we'll share updates, tutorials, and insights about AI-powered development.
