What problem does it solve? Drafting implementation plans by hand often produces inconsistent structures, missing frontmatter, and oversized phases that cannot be reviewed or indexed. This Skill scaffolds plans that match the repository's existing conventions and pass frontmatter validation on the first attempt. ## Core Features & Use Cases - Convention Inference: Scans existing .context/plans/ files to classify sections as core, common, or rare, and follows the majority file-naming convention instead of hardcoding a template. - Graded Frontmatter: Requires effort (S/M/L/TBD), value (HIGH/MEDIUM/LOW), and ordered themes from a closed vocabulary, graded against a rubric rather than guessed. - Phased Decomposition: Splits work into 2-5 phases of 2-5 tasks each, with exit criteria and explicit wave annotations for parallel work. - Use Case: A developer asks to plan an SQS DLQ redrive feature. The Skill gathers goal and scope, drafts a plan with effort: M and value: HIGH, validates the frontmatter with a shell script, and offers to hand off to the plan-review skill. ## Quick Start Ask the agent to create a plan for your next piece of work, describing the goal, scope, and rough implementation steps.
