What problem does it solve?
After an orchestrator finishes an execution round in a multi-agent CCB workflow, someone must decide whether the round passed, needs rework, or requires replanning. This Skill standardizes that judgment by turning orchestrator summaries, worker reports, checker reports, and test evidence into a single machine-readable round result.
Core Features & Use Cases
- Structured verdict output: Produces one of five explicit results: pass, rework_node, partial, replan_required, or global_blocker.
- Evidence-based rules: Requires full verification-contract evidence for pass, and constrains each failure verdict to a precise meaning (bounded node repair, preserved independent branches, out-of-loop replanning, or blocked progress).
- Use Case: In an agentic loop workflow, after the orchestrator returns a round summary with node reports, run this Skill to classify the round as partial so successful independent branches are preserved while dependent branches are marked unfinished.
Quick Start
Verify the latest CCB execution round using the orchestrator summary, worker and checker reports, and command evidence, then output the round result verdict.