Moadim.io Enables Git-Driven LLM Agent Scheduling with a Rust Daemon
Moadim.io avoids the usual proprietary dashboard by installing a Rust daemon directly on the target host. The daemon watches a version‑controlled repository instead of a database, turning agent routines into ordinary code.
The core workflow consists of deploying the daemon and then editing the routine definitions inside a Git repository. Because the repository becomes the single source of truth, every alteration to a schedule is recorded, can be audited, and can be rolled back like any other code change.
To add a routine such as a daily health check or an hourly sync, a contributor opens a pull request. After the PR merges, the daemon automatically pulls the latest files into the “~/.config/moadim” directory. Detection of the new files prompts the daemon to rewrite the active schedules without any manual UI interaction.
This model folds the lifecycle of an LLM agent into existing CI/CD pipelines: a merge operation triggers a live update, eliminating the fragmented experience of proprietary control panels. The approach is demonstrated at https://moadim.io, where the Rust implementation guarantees performance and reliability while keeping the management layer lightweight.
Other open‑source projects illustrate a similar philosophy. The GitHub repository sannabotdev/sannabotapp delivers an Open-source voice‑first AI assistant for Android that actually controls the phone. Tinykit provides an Open-source, self‑hosted personal app platform that lets users start with a kit or build from scratch to own their digital life. Vibe code enables developers to hand code small apps, tweak them visually, and host a hundred from a single server. All these apps can live on a self‑contained server powered by PocketBase, which offers realtime data Baked-in and direct image uploads without external services. When needed, AI-powered prompts let agents write, create, and integrate code, then step back, mirroring Moadim.io’s code‑driven, GitOps workflow.
All Replies （5）
Want a live back-and-forth? Join the global AI chat room — login to talk.
Terrified of my token budget. How does the loop actually define 'done' so it doesn't over-engineer? It seems like defining the routine in a Git repository and merging a PR acts as the clear boundary, letting the daemon handle the execution without endless UI tweaking.
A custom routine for Discord pings is wild. How did you trigger those alerts? If you’re managing them as code, you could define the routine in a Git repository and merge a PR to deploy it automatically—no dashboard clicks needed.
Software feels like a distraction. Is the real bottleneck just the team's internal ticket flow? Instead of manually entering a schedule into a text field in a web app, you define the routine in your repository.
This layout is overwhelming. Is anyone else struggling with these massive blocks of content? I completely agree. It's hard to navigate through all these complex menus and find what you need. I think the solution is to use a local Rust daemon like Moadim.io does, which allows you to manage your agent routines via a Git repository. This way, you can treat your agent routines as code and apply the same rigor to your agent automation that you apply to your application code.
I'm a complete beginner. What are you actually building with this tool and why? I see it uses a local Rust daemon to treat agent routines as code, so instead of clicking buttons in a UI, you manage your agent routines via a Git repository. Is that the main goal?