v1.2.7

This will be the last release of the pipeline with only 3 agents (main, sub, and diagnostic). The pipeline functions but files are not organized on outputs, and steps are not validated to confirm that objects and files produced are validated before the start of the next step. That will happen in v1.3.0, coming soon.

v1.2.4

Okay, now we are cooking. The pipeline has some more stability between failure and restart both on the user and agent side.
This version allows for clean-up and prompt injection between runs.

Future versions will likely add progressively more error handling.

v1.1.4

What's Changed

The pipeline now changed out models to primarily function on qwen3-coder-next and has improved head and sub-agent orchestration and task handling. Pipeline how stores artefacts and progress checkpoints in a mem0 server for reflection.

Future improvements will likely focus on more verbose planning steps prior to any code being generated by a potential agent team in order to create step dependency trees and flow charts with planning validation to confirm the presented scripts will complete the master input prompt with full vision alignment.

Full Changelog: v1.0.0...v1.1.4

v1.0.0

This is the first stable implementation of the pipeline.

Future releases will focus on expanded functionality and ease of setup and use.