Hi HN,

I've been exploring a different problem space from AI agents and LLM frameworks.

Many tools focus on prompts, orchestration, and model capabilities, but production systems often fail because of runtime issues: retries, state loss, failed jobs, crashes, and recovery.

I built ALGOgent Runtime SDK, a lightweight Python toolkit that adds:

- Retry engine - Backoff strategies - State persistence - Checkpoints - Event bus - Logging - Metrics

The goal is simple: make automation scripts and AI workflows more resilient without requiring cloud services, GPUs, or heavy dependencies.

Project page: https://stateflow-dev.github.io/algogent/

I'd appreciate feedback on the concept and whether this is a problem developers actually experience in production.