skills.rest
danielgap
danielgapCOMMUNITY
Explore all skills by this author →
Website𝕏@soydanielgapGitHub

Dual-LLM Review Skill

Runs two parallel LLM reviewers to produce a fused verdict report.

4|Updated Apr 9, 2026
Search Tags:#risk-assessment#adversarial-review#dual-llm#pipeline-evaluation#strategy-build-judge#fusion-report#model-diversity
One-click install
npx skills add https://github.com/danielgap/openclaw-planitor --skill dual-llm-review-skill
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Dual-LLM Review Skill
Source: https://github.com/danielgap/openclaw-planitor/tree/main/skills/dual-llm-review
Command: npx skills add https://github.com/danielgap/openclaw-planitor --skill dual-llm-review-skill

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Dual-LLM Review provides an adversarial evaluation framework that runs two independent LLM reviewers in parallel to mitigate model biases and produce a consolidated verdict for critical pipeline phases.

Core Features & Use Cases

  • Parallel dual-LLM evaluations to surface discrepancies and convergences.
  • Fusion: combine outputs into a single, actionable JUDGE-REPORT style verdict.
  • Phase targeting for Strategy (5), Build (6), and Judge (8) within Planitor workflows.
  • Safe fallback when only one model is available, with explicit caveats in the report.
  • Easy integration with Planitor artifacts to enhance decision quality and risk assessment.

Quick Start

Provide input artifacts for a Planitor phase and run Dual-LLM Review to generate a fused assessment.

Frequently Asked Questions about Dual-LLM Review Skill

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

FAQPage Schema
What is dual-LLM adversarial review and how does it mitigate model bias?▼

Dual-LLM adversarial review mitigates model bias by running two independent LLM reviewers in parallel on plan artifacts, surfacing discrepancies and convergences to produce a consolidated, fused verdict.

How do I run a dual-LLM evaluation on plan artifacts?▼

To run a dual-LLM evaluation, provide input artifacts for a Planitor phase and execute the review to generate a fused assessment, targeting Strategy, Build, or Judge phases within your pipeline.

Can I use a single model for adversarial review if I lack multiple LLM providers?▼

Yes, if only one model is available, the dual-LLM review uses two different prompts or prompts from different providers to preserve cross-checking, explicitly logging this fallback caveat in the final report.

What's the best way to consolidate divergent LLM review outputs into a single verdict?▼

The best way to consolidate divergent outputs is through dual-LLM review fusion, which combines parallel evaluations into a single, actionable JUDGE-REPORT style verdict for robust risk assessment.

Which pipeline phases does the dual-LLM review target for risk assessment?▼

The dual-LLM review targets Strategy (5), Build (6), and Judge (8) phases within Planitor workflows, evaluating plan artifacts to enhance decision quality and risk assessment across these critical stages.

Related Skills

Explore More Product & Management Skills→
danielgapdanielgap
4

pipeline-judge-dual

Merge dual-LLM evaluations into consolidated judge reports for business plans.

Community
Advanced
lossyroblossyrob
39

multi-model-review

Review PRs across multiple AI models and synthesize independent assessments.

Community
Advanced
aura-farmingaura-farming
3

methodology-audit

Audit MEDDPICC compliance across sales pipeline stages using HubSpot data.

Community
Intermediate
akillnessakillness
40

aider-cli-workflow

Automate a reviewable AI-assisted coding loop with pre-commit validation in local Git repositories.

Community
Advanced