Recently I released a free Community version of Dagychu. It is a self-hosted platform for running and managing automation, primarily for Python jobs and pipelines.
We have been working on the project itself for more than 2.5 years. The idea for such a project appeared when, while creating almost any new product, I needed a tool that could handle heavy processes. I wanted to move away from building architecture around cron jobs, and at the same time I really wanted to get the result of jobs by a certain time (deadline), instead of just starting them on a schedule (cron). So initially it was a lightweight orchestrator for deadline-based tasks.
Over time, we needed to solve more complex tasks with higher reliability requirements. I also wanted to be able to look through all automation processes and track what was happening at a particular point in time. This is how Dagychu appeared.
The project developed especially quickly in 2025 with the help of Cursor, which allowed us to rethink the architecture and speed up development several times.
At the moment, Dagychu is convenient because it can be deployed quickly and used to execute Python scripts (but not only Python) without changing their code, while not requiring much additional configuration apart from a YAML description of dependencies and execution parameters.
If I try to summarize and make it more concrete, Dagychu allows you to:
- run jobs locally or in isolation without requiring changes to them (the isolation is provided by Dagychu itself; jobs are executed inside Docker containers when Docker mode is enabled, which is the default)
- work with several projects with different environments at the same time
- combine jobs into a pipeline with dependencies
- use triggers: cron, deadline, webhook, dependency (execution is initiated by completion of the required plan or pipeline)
- keep the history of executions and their states
For example, Dagychu can be used for ETL/ELT, data synchronization, file processing, agent-related workloads (prompt preparation and versioning through changes in input data), or simply long-running processing behind a backend.
Dagychu is not an entirely new product. What I released is a free version based on the core layer of the Enterprise version, which is already used in production processes in different areas, including e-commerce and AI/software.
Installation usually takes 1-2 minutes, after which you can add and run your own scripts. All secrets, code and logs stay on the host where Dagychu is running.
The application source code itself is not published in the repository. The Community version is distributed as Docker images available for use.
AI was used during Dagychu development for writing code, tests and reviews. But Dagychu itself is not an AI application and does not require users to provide API keys for LLMs.
Now the product is entering a new stage as a self-hosted product, where we allow users to install and operate it themselves. So my main interest right now is to understand how viable the product is without the involvement of Dagychu developers and outside the environments where it is currently being used.
So if you have a project where the backend needs to execute fairly heavy processes, primarily related to Python code, if you run cron jobs but would like your automation to be observable, controllable and recoverable, I would suggest trying to run at least one process with Dagychu and giving feedback: how easy it was and what problems or limitations you encountered.
Recently I released a free Community version of Dagychu. It is a self-hosted platform for running and managing automation, primarily for Python jobs and pipelines.
We have been working on the project itself for more than 2.5 years. The idea for such a project appeared when, while creating almost any new product, I needed a tool that could handle heavy processes. I wanted to move away from building architecture around cron jobs, and at the same time I really wanted to get the result of jobs by a certain time (deadline), instead of just starting them on a schedule (cron). So initially it was a lightweight orchestrator for deadline-based tasks. Over time, we needed to solve more complex tasks with higher reliability requirements. I also wanted to be able to look through all automation processes and track what was happening at a particular point in time. This is how Dagychu appeared.
The project developed especially quickly in 2025 with the help of Cursor, which allowed us to rethink the architecture and speed up development several times. At the moment, Dagychu is convenient because it can be deployed quickly and used to execute Python scripts (but not only Python) without changing their code, while not requiring much additional configuration apart from a YAML description of dependencies and execution parameters.
If I try to summarize and make it more concrete, Dagychu allows you to:
- run jobs locally or in isolation without requiring changes to them (the isolation is provided by Dagychu itself; jobs are executed inside Docker containers when Docker mode is enabled, which is the default) - work with several projects with different environments at the same time - combine jobs into a pipeline with dependencies - use triggers: cron, deadline, webhook, dependency (execution is initiated by completion of the required plan or pipeline) - keep the history of executions and their states
For example, Dagychu can be used for ETL/ELT, data synchronization, file processing, agent-related workloads (prompt preparation and versioning through changes in input data), or simply long-running processing behind a backend.
Dagychu is not an entirely new product. What I released is a free version based on the core layer of the Enterprise version, which is already used in production processes in different areas, including e-commerce and AI/software.
Installation usually takes 1-2 minutes, after which you can add and run your own scripts. All secrets, code and logs stay on the host where Dagychu is running.
The application source code itself is not published in the repository. The Community version is distributed as Docker images available for use. AI was used during Dagychu development for writing code, tests and reviews. But Dagychu itself is not an AI application and does not require users to provide API keys for LLMs.
Now the product is entering a new stage as a self-hosted product, where we allow users to install and operate it themselves. So my main interest right now is to understand how viable the product is without the involvement of Dagychu developers and outside the environments where it is currently being used.
So if you have a project where the backend needs to execute fairly heavy processes, primarily related to Python code, if you run cron jobs but would like your automation to be observable, controllable and recoverable, I would suggest trying to run at least one process with Dagychu and giving feedback: how easy it was and what problems or limitations you encountered.