halfmoon-mind

halfmoon-mind/fusion-council

Plugin Claude Code

Deterministic multi-model planning & review councils for Claude Code (Claude role panel + GPT-5.6 Sol).

halfmoon-mind/fusion-council

Plugin Claude Code

Multi-model planning & review councils for Claude Code: fusion-plan runs a Claude role panel (architect/skeptic/test/maintainer) plus GPT-5.6 Sol via the local Codex CLI; fusion-review runs a bench-validated duo (one generalist Claude seat + GPT-5.6 Sol; role panel via args.roles) — each adjudicated and written up by…

halfmoon-mind/fusion-council

Agent

Use before planning non-trivial changes to map relevant files, tests, existing patterns, and constraints. Do not propose solutions.

halfmoon-mind/fusion-council

Agent

Use for architecture decisions, refactors, and non-trivial implementation planning. Produces the smallest safe implementation plan.

halfmoon-mind/fusion-council

Agent

Use to check whether a plan fits the codebase style, ownership boundaries, maintainability, migration needs, and public API constraints.

halfmoon-mind/fusion-council

Agent

Use to challenge a proposed plan, find overengineering, hidden coupling, simpler alternatives, and likely failure modes.

halfmoon-mind/fusion-council

Agent

Use to define verification criteria before implementation or review whether a plan has enough tests.

halfmoon-mind/fusion-council

Skill Claude CodeCodex

Run the deterministic multi-model planning council (Claude role panel + GPT-5.6 Sol) before a non-trivial change, then implement only after the user approves.

halfmoon-mind/fusion-council

Skill Claude CodeCodex

Run the deterministic multi-model review council (generalist Claude seat + GPT-5.6 Sol; args.roles restores the Claude role panel) over the current working-tree diff after implementation.