Evaluate and refine implementation plans through parallel multi-agent review.
npx skills add https://github.com/ian97531/goodplan --skill refine-plan-ian97531
Please help me install this Agent Skill. Skill: refine-plan Source: https://github.com/ian97531/goodplan/tree/main/skills/refine-plan Command: npx skills add https://github.com/ian97531/goodplan --skill refine-plan-ian97531
Iteratively improve a plan document by spawning specialized review sub-agents in parallel, synthesizing their feedback, performing research when needed, and repeating until the plan is ready for implementation.
Run /refine-plan <path-to-plan> to start refining the plan.
High-intent search queries and answers about installing and using this skill.
Iterative plan refinement spawns parallel review sub-agents to synthesize feedback and edit your plan. It loads repository context and decisions, running up to 12 refinement loops until the plan reaches production-ready status.
Reviewing a software architecture plan is best handled by parallel domain-specialist agents that assess structure and context. They synthesize feedback into targeted edits, preserving the original plan as a source of truth.
Automated plan refinement works across domains including machine learning, web, and systems. It applies multi-agent evaluation and research triggers to iteratively improve your ML strategy document until it meets explicit exit criteria.
The plan refinement loop triggers research automatically when reviewers identify gaps. It synthesizes the findings into feedback, applying targeted edits to the plan document before the next evaluation iteration begins.
Automated multi-agent plan refinement enforces a strict 12-iteration cap and requires explicit exit criteria to stop. It transitions the document to a refined state, preserving the original plan separately as the source of truth.
Loading repository decisions and conventions happens automatically during context initialization for plan evaluation. The system grounds the refinement process by reading active epics and project decisions before spawning review agents.
SafphereGuide collaborative document creation through context gathering, refinement, and reader testing.
xiasong0501Compare episode script versions to diagnose pacing and emotional impact, then generate a prioritized rewrite plan.
songlairuiNormalize raw project state into a standardized handoff context for AI agents.
scooter-lacroixOrchestrate multi-agent code review with parallel specialized agents and synthesized feedback.
