Skales v12.9.25 - Backbone
Backbone is the release where Skales got one engine. A chat you type, a task you scheduled, a job on a timer, a plugin, a webhook, a Discord message, a Telegram or WhatsApp turn and the desktop Buddy used to run on different loops with different rules; now they all run on the same one, with the same guards, the same budget and the same record of what happened. Close Skales in the middle of a run, update it, let it crash: the run comes back as an unfinished run, says how far it got, and one Continue picks it up at that step. Ask for something big and Skales can hand parts of it to sub-agents, each in its own checkout of your repository, and collect the results. Under every reply sits a context meter that names what filled the window: the system prompt, the tool definitions, your skills, your memory and the conversation itself.
We've just released Skales v12.9.25, a new version of our private, local-first AI desktop agent. This update focuses on making Skales more reliable and predictable in how it handles information and actions.
Memory Now Understands Sensitive Data
Skales will no longer save sensitive information like card numbers, IBANs, API keys, or passwords directly into its memory. Instead, it will clearly state that it hasn't saved it and direct you to the appropriate place, like the Settings menu for provider keys. Ordinary facts, like budgets or version numbers, are still remembered as before. For genuinely sensitive data you wish to retain, there's a new switch in Settings, Memory. Automatically distilled facts are also now limited to a single line to prevent long paragraphs from cluttering your prompts.
Tools Only Act on What They See
Before Skales performs an action like sending an email or changing a file, it now checks that the item it's supposed to interact with was actually present in the conversation. This prevents the agent from acting on identifiers generated out of thin air. If an unrecognized identifier is encountered, Skales will stop, inform you which lookup would resolve it, and then continue. This check happens before you're asked to confirm anything, ensuring confirmations are always based on real data.
A Unified Engine for All Operations
All parts of Skales, from chats and scheduled tasks to plugins and messaging integrations, now run on the same agent loop. This means a consistent approach to how requests are handled and managed. Your set budget now applies everywhere, including for Buddy messages, scheduled tasks, and messages sent via Telegram or WhatsApp. You can also see and stop long-running tasks from the run list.
Runs Now Survive Restarts
If Skales closes, updates, or crashes while a task is in progress, the run will no longer be lost. It will reappear in the conversation as an unfinished run, indicating its progress, and can be resumed from that point. Interrupted autonomous runs will also pick themselves up on the next start.
New here? Skales is the private AI agent you install like an app - free for personal use, any model, everything stays on your machine.
Download Skales