gsd-verifier

Validates implemented code against phase goals using three-level artifact and wiring checks.

1|Updated Sep 3, 2026
One-click install
npx skills add https://github.com/amanpal3/SKILLs --skill gsd-verifier-amanpal3
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: gsd-verifier
Source: https://github.com/amanpal3/SKILLs/tree/main/.agent/plugins/gsd/skills/gsd-verifier
Command: npx skills add https://github.com/amanpal3/SKILLs --skill gsd-verifier-amanpal3

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Teams often mark development phases as complete based on claims in summary documents, only to discover later that code is stubbed, unwired, or missing. This Skill verifies implemented work against stated phase goals using empirical evidence instead of trusting completion claims. ## Core Features & Use Cases - Three-Level Artifact Verification: Checks that required files exist, contain substantive implementations (not stubs or placeholders), and are actually wired into the application. - Key Link and Wiring Validation: Confirms critical connections such as component-to-API calls, API-to-database queries, and form-to-handler bindings. - Structured Gap Reporting: Produces a VERIFICATION.md with status (passed, gaps_found, human_needed), a verification score, anti-pattern findings, and YAML-structured gaps consumable by planning workflows. - Use Case: After a coding agent finishes a chat feature phase, run verification to confirm Chat.tsx fetches from /api/chat, renders messages from state, and contains no TODO stubs before approving the phase. ## Quick Start Verify phase 3 of my project against its roadmap goal and generate a VERIFICATION.md report with any gaps found.

Frequently Asked Questions about gsd-verifier

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

FAQPage Schema
How do I verify that implemented code actually meets requirements?▼

Define must-haves as observable truths, required artifacts, and key links, then check each artifact at three levels: existence, substantive implementation, and wiring. The verifier produces a VERIFICATION.md with per-item status and evidence.

How to detect stub or placeholder code in a codebase?▼

Scan for patterns like TODO, FIXME, placeholder text, empty returns (null, {}, []), handlers that only call preventDefault, and API routes returning static responses without database queries. Findings are categorized as blockers, warnings, or info.

What is the difference between re-verification and initial verification?▼

If a previous VERIFICATION.md exists with gaps, re-verification mode runs full three-level checks only on previously failed items and quick regression checks on passed items. Initial mode establishes must-haves from PLAN frontmatter or derives them from the phase goal.

Can automated verification check visual appearance and user flows?▼

No. Visual layout, real-time behavior like WebSockets, external service integration, and error message clarity require human verification. The verifier flags these items in a dedicated human verification section with test steps and expected outcomes.

What happens when verification finds gaps in the implementation?▼

The status is set to gaps_found and gaps are structured in YAML with the failed truth, reason, affected artifacts, and missing pieces. This output is designed to feed directly into a planning step such as /plan --gaps for remediation.