Moadim.io launches a universal scheduler for AI agents

The engine automates loop creation, removing

Moadim.io presents a loop engine that lets developers schedule AI agents to run continuously on code repositories. The platform supports multiple models, integrates with Git, and isolates each execution in a watchdog‑protected workbench. It runs Claude, Codex, Hermes, or Pi on every tick, or queues NanoClaw tasks for asynchronous processing.

The scheduler treats a loop as a prompt, a schedule, and an agent. Each tick triggers a model call against the repository, while a watchdog monitors runs to prevent runaway processes. Developers define loops once and let the engine handle repeated execution without manual prompting. The loop engine also supports queued tasks for background processing.

Isolated workbenches protect each agent run Each agent runs inside a sandboxed workbench. A watchdog terminates runs that exceed time limits or consume excessive resources. This isolation prevents interference between concurrent tasks. Agents can be restarted automatically if they fail.

The engine automates loop creation, removing the need for hand‑crafted prompts. As the developer explains, „Loop engineering, not prompting by hand.”

How does Moadim ensure security during agent

Can a single scheduler handle all major AI models? Moadim supports Claude, Codex, Hermes, and Pi without additional configuration. The scheduler selects the appropriate model for each tick, ensuring consistent performance across services.

Developers gain reliable, repeatable execution of AI tasks without constant manual oversight. The approach could streamline agent workflows in research, automation, and production environments. Future updates may add more model adapters and richer monitoring tools. Adoption could accelerate AI research cycles and reduce operational overhead.

How does Moadim ensure security during agent execution? Each loop runs in an isolated workbench with a watchdog that monitors resource use and terminates unsafe behavior. This sandbox prevents agents from accessing the host system or other workloads.

Which programming languages are required to use the scheduler? Only a Rust‑based CLI is needed; the command cargo install — installs the tool. Developers then define loops in a simple text format.

Can the scheduler be integrated with existing CI pipelines? Yes, the engine can be triggered as a step in CI jobs, allowing continuous testing of AI‑driven code changes. Integration is straightforward via command‑line invocation.