external-consensus

Synthesize a consensus plan from multi-agent debate reports by combining bold, critique, and reducer perspectives into a single actionable implementation file. An external AI reviewer processes the three agent reports to produce a structured, ready-to-use plan.

Multi-agent consensus planning is the process of merging conflicting AI debate outputs into an implementation-grade plan. You need it when a multi-agent workflow produces divergent bold, critique, and reducer reports that require external review to resolve.

External-consensus supports both Claude Opus and Codex. It prefers Codex for deep reasoning with web research capabilities, but Claude Opus is available as a fallback to synthesize the debate reports into a final plan.

Generate an implementation plan from three AI perspectives by invoking the external-consensus script with the bold, critique, and reducer agent reports. The external reviewer combines these views to output a concrete consensus file ready for GitHub issue creation.

A limitation of using an external AI reviewer for multi-agent debate synthesis is the strict dependency on three distinct agent debate reports. Without the bold, critique, and reducer perspectives as input, the script cannot produce an implementation-grade consensus plan.