GSD Plan Checker

Validates PLAN.md files against six dimensions before execution to catch blockers early.

Updated Mar 11, 2026
One-click install
npx skills add https://github.com/apsulli/gsd-extended --skill gsd-plan-checker-apsulli
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: GSD Plan Checker
Source: https://github.com/apsulli/gsd-extended/tree/main/.agent/skills/plan-checker
Command: npx skills add https://github.com/apsulli/gsd-extended --skill gsd-plan-checker-apsulli

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Plans written for AI-assisted execution often contain hidden flaws—uncovered requirements, missing verification steps, circular dependencies, or oversized scope—that only surface as failures mid-execution. This Skill inspects PLAN.md files before execution begins, catching blockers and quality risks while they are still cheap to fix. ## Core Features & Use Cases - Six-Dimension Validation: Checks requirement coverage, task completeness, dependency correctness, key links between artifacts, scope sanity, and verification derivation. - Severity-Graded Issues: Classifies findings as blocker, warning, or info with structured YAML issue reports including fix hints. - Pass/Fail Quality Gate: Produces a clear PASSED or ISSUES_FOUND status that gates the transition from planning to execution. - Use Case: After generating plans for a phase with /plan, run the plan checker to discover that a logout requirement has no covering task and that two plans have a circular dependency—then fix both before any code is written. ## Quick Start Ask the AI to run the GSD plan checker on the current phase plans and report any blockers or warnings before execution.

Frequently Asked Questions about GSD Plan Checker

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I validate implementation plans before executing them?▼

Run the plan checker after planning completes and before execution starts. It reads ROADMAP.md and all PLAN.md files for the phase, checks six validation dimensions, and reports blockers and warnings with suggested fixes.

What makes a task plan complete enough to execute?▼

Each auto task needs Files, Action, Verify, and Done elements. Missing verification commands or acceptance criteria are flagged as blockers because completion cannot be confirmed without them.

How are plan dependencies and waves checked for cycles?▼

The checker parses depends_on from each plan's frontmatter, builds a dependency graph, and flags circular references, missing plan references, and wave assignments inconsistent with the dependency chain.

When should I split a plan into multiple plans?▼

Split when a plan exceeds 4 tasks or 10 file modifications. Plans with 5+ tasks or 15+ files are flagged as blockers because execution quality degrades and context budgets are exceeded.

What is the difference between a blocker and a warning in plan validation?▼

A blocker will cause execution failure and must be fixed before running /execute. A warning indicates a quality or efficiency risk that should be addressed but does not prevent proceeding.