plan-checker_skill

Overview

This skill validates plans before execution to catch issues that would cause failures or reduce quality. It inspects plan structure, task details, dependencies, scope, and traceability to phase goals. The checker surfaces blockers and warnings and provides actionable fix hints so plans are safe to execute.

How this skill works

The checker parses each plan and its metadata, extracts phase goals, tasks, must-haves, and declared dependencies. It runs six focused validations: requirement coverage, task completeness, dependency correctness, artifact wiring, scope sanity, and verification derivation. Results are classified by severity and returned as a concise report with suggested fixes.

When to use it

• Immediately after plan authoring and before execution
• After any plan edits or when merging plan changes
• As a gate during CI or a manual review workflow
• When handoffs occur between planners and implementers

Best practices

• Keep tasks small (2–3 per plan) and explicit: include files, action, verify, done
• Declare only valid, earlier dependencies and avoid cycles
• Trace every must-have back to the phase goal; avoid unrelated must-haves
• Wire artifacts deliberately: mention API calls, handlers, and state rendering where relevant
• Split plans that exceed task/file thresholds to preserve clarity

Example use cases

• Validate a phase plan to ensure every requirement has at least one concrete task
• Detect missing verification steps so testers can confirm completion
• Catch circular or future-facing dependencies before execution
• Find components created but never wired to APIs or handlers
• Flag oversized plans that should be split for manageable execution