← All selected work

AI & automation / Engineering project

Multi-agent orchestratorGive AI work a deliberate workflow.

A planner-to-worker orchestration system for coordinating frontend and backend generation with configurable model providers.

01 / THE PROBLEM

What needed to change.

Generating software across several models requires more than a single prompt: tasks need boundaries, outputs need validation, and failures need to remain visible.

02 / THE APPROACH

The work behind the product.

  • Structured planning and parallel frontend/backend workers around explicit file tasks.
  • Supported provider configuration, role-specific models and bounded retries.
  • Separated generated artifacts from filesystem writes, with workspace boundary checks and regression tests.

03 / THE OUTCOME

Where the work stands.

A configurable orchestration foundation and a documented activation protocol for agents joining collaborative projects.

What comes next

Artifact validation does not establish generated-code correctness. Test execution and product acceptance remain separate steps.

EXPLORE ANOTHER PROJECTCRDLedger × BMONI ↗