What problem does it solve? Choosing between multiple possible solutions and turning the decision into a concrete implementation plan is slow and error-prone. This Skill structures that decision process inside the Pi coding agent, comparing practical candidates against the ideal end state and producing a step-by-step plan. ## Core Features & Use Cases - Structured Candidate Comparison: Records two to four distinct practical solutions with rationale, trade-offs, and rejection reasons, plus a separate ideal (Holy grail) option. - Workflow Integration: Starts the built-in autoplan workflow in pi-workflows with a structured input containing problem, scope, constraints, and optional previous plan and new evidence. - Plan Revision Support: Revises an existing plan when new evidence appears, preserving the previous plan and recording why it was kept or rejected. - Use Case: You need to decide how to add a timeout fallback to a repository's public API. Run autoplan to compare candidate designs, select the best in-scope option, and receive a detailed implementation plan with verification steps for each change. ## Quick Start Ask the agent to run autoplan to choose the best solution for your stated problem and write an implementation plan for the current repository.