kiro-verify-completion

Verify completion claims using fresh command outputs and exit codes.

Updated Apr 17, 2026
One-click install
npx skills add https://github.com/BarryYin/Hermes-sdd --skill kiro-verify-completion-barryyin
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kiro-verify-completion
Source: https://github.com/BarryYin/Hermes-sdd/tree/main/skills/kiro-verify-completion
Command: npx skills add https://github.com/BarryYin/Hermes-sdd --skill kiro-verify-completion-barryyin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents false or premature completion claims by requiring fresh, scoped, and verifiable evidence before accepting that a task, fix, test/build, or feature is complete.

Core Features & Use Cases

  • Fresh Evidence Requirement: Enforce collection of current command outputs, exit codes, and validation artifacts rather than relying on prior reports.
  • Claim-Specific Gates: Apply tailored checks for TASK, FIX, TEST_OR_BUILD, and FEATURE_GO claims, including requirement coverage and integration checks for feature-level validation.
  • Escalation Paths: Return clear outcomes (VERIFIED, NOT_VERIFIED, MANUAL_VERIFY_REQUIRED) with gaps and next actions when verification cannot be completed.
  • Use Case: When an autonomous subagent reports a fix, use this Skill to run the canonical reproduce-and-validate commands, inspect results, and decide whether to accept the fix or require manual follow-up.

Quick Start

Ask the skill to verify the claim by providing the exact claim text, claim type, the canonical validation commands, and the latest command outputs and exit codes.

Frequently Asked Questions about kiro-verify-completion

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

FAQPage Schema
How do I verify task completion claims with fresh evidence in spec-driven development?

Verify completion claims by requiring fresh command outputs, exit codes, and validation artifacts for TASK, FIX, TEST_OR_BUILD, and FEATURE_GO claims, returning VERIFIED, NOT_VERIFIED, or MANUAL_VERIFY_REQUIRED outcomes.

How do I prevent false completion reports from autonomous agents?

Prevent false completion reports by applying claim-specific gates that enforce current command outputs and exit codes rather than relying on prior reports, ensuring autonomous agents provide verifiable evidence for every fix or task.

What's the best way to validate feature completion against requirements and design alignment?

Validate feature completion by applying FEATURE_GO claim checks that require requirement coverage, design alignment verification, and integration evidence, returning clear outcomes with gaps and next actions when verification cannot be completed.

Can I use spec-driven verification for both human reports and autonomous agent claims?

Yes, spec-driven verification applies to both autonomous agent and human reports during spec-driven development workflows, using canonical validation commands and fresh evidence to determine whether to accept or require manual follow-up.

When should I expect a MANUAL_VERIFY_REQUIRED outcome during completion verification?

Expect MANUAL_VERIFY_REQUIRED when verification cannot be completed using available fresh evidence, command outputs, or exit codes, providing clear gaps and next actions for manual follow-up inspection.

What claim types does spec-driven completion verification support?

Spec-driven completion verification supports four claim types: TASK, FIX, TEST_OR_BUILD, and FEATURE_GO, each with tailored checks including requirement coverage and integration checks for feature-level validation.