parallel-agent-research-verification - SKILL.md Agent Skill

name: parallel-agent-research-verification description: Uses parallel agent workflows to prevent hallucinations in research tasks by implementing concurrent verification loops. Triggers when AI agents produce incorrect information despite having sources, or when research quality degrades due to repeated manual corrections. Apply for fact-checking, source verification, and research accuracy improvement. source: Transcript File Import + https://www.youtube.com/watch?v=DEMO generated_by: SkillJack (skilljacked.com)

Parallel Agent Research Verification

Prevent AI hallucinations in research by deploying parallel verification agents that cross-check facts and sources concurrently.

Quick Start

When your research agent produces incorrect information despite having sources, deploy a parallel verification agent to fact-check claims in real-time while the primary agent continues working.

Core Workflow

Trigger: Research Quality Degradation

When you notice:

Agent hallucinating facts despite provided sources
Repeated manual corrections needed
Research feeling "pointless" due to constant fixes

Setup Process

Identify Parallelizable Tasks - Separate research generation from verification
Deploy Primary Research Agent - Handles main research and comparison tasks
Launch Verification Agent in Parallel - Cross-checks facts and sources simultaneously
Configure Auto-Detection Override - Even though Claude detects parallel opportunities, explicitly structure parallel workflows

Execution Pattern

Research agent generates content
Verification agent validates claims against sources concurrently
Both agents communicate findings
Corrections happen in real-time rather than post-generation

Techniques

Parallelization Structure

ALWAYS separate fact-generation from fact-checking into parallel streams
ALWAYS provide same sources to both research and verification agents
ALWAYS configure explicit parallel workflows even when Claude auto-detects opportunities

Verification Loops

Run verification continuously during research, not after completion
Cross-reference multiple sources per claim
Flag discrepancies immediately for human review

Anti-Patterns

Sequential Verification

NEVER wait until research completion to verify facts
NEVER rely solely on single-agent self-correction for factual accuracy
NEVER ignore repeated hallucinations without implementing parallel verification

Resource Mismanagement

NEVER skip parallelization due to token cost concerns when accuracy is critical
NEVER assume Claude's automatic parallel detection covers all verification needs

Edge Cases & Error Handling

Source Conflicts

When verification agent finds contradictory sources, escalate to human review
Maintain source hierarchy (primary vs secondary sources)
Document conflicting information rather than choosing arbitrarily

Token Usage Management

Monitor increased token consumption from parallel agents
Balance speed gains against cost implications
Consider parallel workflows essential investment for accuracy-critical research

Bundled Resources Plan

scripts/parallel_research_setup.py - Template for deploying research + verification agent pairs
references/verification_prompts.md - Proven prompts for fact-checking agents
assets/source_hierarchy_templates.json - Source prioritization configurations