If there is one thing that universally ruins a developer's flow state, it’s trying to accurately log time.
When you spend your day jumping frantically between a C# backend, complex SQL migrations, and a JavaScript frontend, your "schedule" is rarely linear. You get interrupted, you context-switch, and by 5:00 PM, trying to piece together exactly how long a specific task took is a guessing game.
I originally built SheepCat as a local-first, zero-cloud AI scratchpad to handle this context-switching for me. You drop messy notes in, and the local AI formats them. But as the community started using it, two massive workflow bottlenecks became obvious: timing accuracy and historical summaries.
Today, with the release of SheepCat v01.04.00, we are fixing both.
⏱️ The Timing Fix: Chain-Based & Fixed Durations
Previously, tracking the exact duration of an interrupted task was clunky. If you went down a rabbit hole debugging an auth loop, the timing could easily get skewed.
In v1.4, we’ve introduced chain-based task timing and special tasks with fixed durations.
What does this mean for your workflow? It means SheepCat now understands the fluid nature of dev work. Instead of rigid start/stop timers that you inevitably forget to click, the system better handles chained events and allows you to set fixed durations for specific, repetitive tasks (like that 15 minute daily standup). It reduces the cognitive load of micro managing the clock, keeping the focus entirely on the code.
⏪ The Time Machine: Rerunning Historical Summaries
Because SheepCat uses your local LLM (via Ollama) to generate your end-of-day project updates, we ran into a practical reality of working with AI: sometimes, you just want a do-over.
Maybe you dropped a crucial ticket number into yesterday's log after the daily summary ran. Or maybe you tweaked your local Ollama model prompt and want to see how it formats last week's messy notes.
In this release, we added the Summary History page.
You can now navigate back to any day in your history and trigger a completely fresh rerun of your full-day summary. No data is locked in. If you missed a task yesterday, just add it today, go to the history page, and let the local AI regenerate a perfectly clean update for your PMs.
Protecting the Flow State
The entire philosophy behind SheepCat is cognitive ergonomics. Your tracking tools should never be a source of stress. They should sit quietly in the background, keeping your proprietary code completely offline, and doing the administrative heavy lifting for you.
If you are tired of cloud-based tracking tools breaking your focus (or leaking your data), you can check out the new v1.4 release on GitHub here:
👉 SheepCat-TrackingMyWork Release v01.04.00
Would love to hear how the local AI summaries are working for your stacks!
Top comments (0)