gormes-delivery-loop

To orchestrate a delivery loop with TDD validation, this Skill coordinates a bounded iterative pipeline that converts research findings into actionable work while enforcing parity evidence, gate checks, and commit discipline. It applies architecture review, planner shaping, and TDD-driven building within structured iteration budgets.

A bounded in-agent delivery loop is a structured iterative pipeline for converting research into validated, builder-ready work within Hermes/Honcho contracts. You need this process when coordinating architecture review, planner shaping, parity evidence, and TDD-driven building to ensure safe, repeatable progress with defined stop conditions.

To enforce parity evidence and branch constraints during a delivery loop, the pipeline applies evidence quality guards, gate checks, and safe stop conditions across architecture candidate, planner, builder, and validation stages to ensure structured iteration budgets and commit discipline are satisfied.

This delivery loop orchestration is designed specifically for guided skill development within Hermes/Honcho-integrated environments. It applies bounded iterative pipelines, branch constraints, and evidence quality guards tailored to those contract structures, making it unsuitable for environments lacking that framework.

Limitations of this bounded delivery loop include strict iteration budgets, enforced branch constraints, and defined stop conditions that restrict open-ended exploration. The pipeline requires passing gate checks and parity evidence guards at each stage, which may halt progress if evidence quality is insufficient.

The best way to start a bounded delivery loop is to initiate it using your current project context with a 1-iteration budget. The pipeline then processes research findings through architecture review, planner shaping, parity checks, TDD-driven building, and validation to produce actionable builder-ready work.