skills.rest
michael-conrad
michael-conradCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

verification

Verify text, vision, and audio claims against evidence with PASS/FAIL/UNVERIFIED results.

Updated Apr 24, 2026
Search Tags:#qa#verification#audit#multimodal#evidence#claims
One-click install
npx skills add https://github.com/michael-conrad/.opencode --skill verification-michael-conrad
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: verification
Source: https://github.com/michael-conrad/.opencode/tree/main/skills/verification
Command: npx skills add https://github.com/michael-conrad/.opencode --skill verification-michael-conrad

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Verifies claims against evidence across modalities and returns PASS/FAIL/UNVERIFIED with artifacts, enabling clear, auditable verdicts for every claim.

Core Features & Use Cases

  • Cross-modal verification: routes text, vision, and audio claims to the appropriate model via multimodal-dispatch.
  • Per-claim results: returns a structured ClaimResult with status, evidence, artifacts, and model metadata.
  • Integrity enforcement: guarantees FAIL cannot be downgraded to PASS and supports escalation when verification fails.
  • Use Case: validate a user-reported issue against logs and screenshots to produce a definitive verdict.

Quick Start

Verify the provided claims against the supplied evidence using the multimodal-dispatch workflow.

Frequently Asked Questions about verification

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

FAQPage Schema
How do I verify claims against evidence across text, vision, and audio?▼

Cross-modal claim verification routes text, vision, and audio inputs to the appropriate model and returns a structured PASS, FAIL, or UNVERIFIED verdict with supporting artifacts for each claim.

What is structured cross-modal claim verification and when do I need it?▼

Structured cross-modal claim verification evaluates assertions against multimodal evidence to produce auditable verdicts. It is needed when validating issues against logs and screenshots to reach a definitive, reviewable conclusion.

How do I get per-claim verification results with evidence and model metadata?▼

Per-claim verification results are returned as a structured ClaimResult object containing the status, supporting evidence, artifacts, and model metadata for each evaluated assertion across all modalities.

Does claim verification guarantee that a failed result cannot be downgraded to pass?▼

Strict integrity enforcement guarantees that a FAIL verdict cannot be downgraded to PASS, ensuring honest verification outcomes and automatically escalating failures for developer review.

Can I audit multimodal claims using screenshots and logs together?▼

Auditing multimodal claims with screenshots and logs routes each evidence type to the appropriate model via multimodal-dispatch, validating user-reported issues against combined inputs to produce a definitive verdict.

Related Skills

Explore More Data & Analytics Skills→
michael-conradmichael-conrad

verification-before-completion

Validate success criteria and collect artifacts before task completion.

Community
Advanced
m2ai-portfoliom2ai-portfolio
4

build-spec-generator

Generate a deployment-ready build spec from a workflow verdict.

Official
Intermediate
Owl-ListenerOwl-Listener
157

multimodal-orchestration

Coordinate multimodal AI interactions across text, image, and voice.

Community
Advanced
ya-wayya-way

bio-single-cell-multimodal-integration

Integrate multi-modal single-cell RNA, protein, and ATAC data into unified representations.

Community
Advanced