plan-coordinator

What problem does it solve?

Projects often store implementation plans as scattered markdown files without an actionable, assignable execution plan. This Skill reads those plans, extracts and normalizes every step, assigns specialized executor agents, identifies parallel groups and bundles, and writes a persistent progress file so teams can approve and delegate work reliably.

Core Features & Use Cases

• Execution Map Generation: Parse plan steps (number, title, description, dependencies, inputs, outputs, validation, effort) and produce a group-oriented map that assigns agents and marks parallelism or bundles.
• Agent Discovery & Assignment: Read .claude/agents/*.md metadata to match steps to the most suitable executor, with sensible fallbacks to general-executor.
• Dependency Resolution & Bundling: Resolve inputs and prior outputs, flag skipped or completed steps, and apply bundling rules to balance context budget and retry boundaries.
• Progress File Persistence: Write a full progress-<plan-name>.md containing the Execution Map, recovery state, wave logs, and per-step context blocks for main-context delegation.
• Use Case: Turn a project's docs/plans/<topic>/*-plan.md into an approved, delegation-ready Execution Map that the main context can use to launch executor agent waves.

Quick Start

Parse the plan in docs/plans/<topic> and generate a complete Execution Map with agent assignments and a progress-<plan-name>.md ready for user approval.