Research

Customization

Before executing, check for user customizations at: ~/.opencode/PAI/USER/SKILLCUSTOMIZATIONS/Research/

If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.

Research Skill

Comprehensive research, analysis, and content extraction system.

MANDATORY: URL Verification

READ: UrlVerificationProtocol.md - Every URL must be verified before delivery.

Research agents hallucinate URLs. A single broken link is a catastrophic failure.

Voice Notification

When executing a workflow, do BOTH:

1. Send voice notification:

   curl -s -X POST http://localhost:8888/notify \
     -H "Content-Type: application/json" \
     -d '{"message": "Running the WORKFLOWNAME workflow from the Research skill"}' \
     > /dev/null 2>&1 &
   

2. Output text notification:

   Running the **WorkflowName** workflow from the **Research** skill...
   

Workflow Routing

Route to the appropriate workflow based on the request.

CRITICAL: For due diligence, company/person background checks, or vetting -> INVOKE OSINT SKILL INSTEAD

Research Modes (Primary Workflows)

• Quick research - DEFAULT (1 Claude, FREE) → Workflows/QuickResearch.md
• Standard research (3 agents: Claude + Gemini + Perplexity Sonar, ~$0.01) → Workflows/StandardResearch.md
  Trigger: "standard research", "thorough research", "research X from multiple angles"
• Extensive research (4-5 providers, flexible angles, ~$0.10-0.50) → Workflows/ExtensiveResearch.md Trigger: "extensive research" - REQUIRES USER CONFIRMATION

Controller & Constraints

• Research with enforced recursion limits → ResearchController.md
  READ FIRST: When spawning parallel research agents to prevent Matryoshka Paradox

Deep Content Analysis

• Extract alpha / deep analysis / highest-alpha insights -> Workflows/ExtractAlpha.md

Content Retrieval

• Difficulty accessing content (CAPTCHA, bot detection, blocking) -> Workflows/Retrieve.md
• YouTube URL extraction (use fabric -y URL immediately) -> Workflows/YoutubeExtraction.md
• Web scraping -> Workflows/WebScraping.md

Specific Research Types

• Claude WebSearch only (free, no API keys) -> Workflows/ClaudeResearch.md
• Perplexity API research (use Quick for single-agent) -> Workflows/QuickResearch.md
• Interview preparation (Tyler Cowen style) -> Workflows/InterviewResearch.md
• AI trends analysis -> Workflows/AnalyzeAiTrends.md

Fabric Pattern Processing

• Use Fabric patterns (242+ specialized prompts) -> Workflows/Fabric.md

Content Enhancement

• Enhance/improve content -> Workflows/Enhance.md
• Extract knowledge from content -> Workflows/ExtractKnowledge.md

Quick Reference

READ: QuickReference.md for detailed examples and mode comparison.

Integration

Feeds Into

• blogging - Research for blog posts
• newsletter - Research for newsletters
• xpost - Create posts from research

Uses

• be-creative - deep thinking for extract alpha
• OSINT - MANDATORY for company/people comprehensive research
• BrightData MCP - CAPTCHA solving, advanced scraping
• Apify MCP - RAG browser, specialized site scrapers

File Organization

Scratch (temporary work artifacts): ~/.opencode/MEMORY/WORK/{current_work}/scratch/

• Read ~/.opencode/MEMORY/STATE/current-work.json to get the work_dir value
• All iterative work artifacts go in the current work item's scratch/ subdirectory
• This ties research artifacts to the work item for learning and context

History (permanent): ~/.opencode/History/research/YYYY-MM/YYYY-MM-DD_[topic]/

⚠️ MANDATORY: Research Agent Constraints

CRITICAL RULES to prevent Matryoshka Paradox (recursive agent spawning):

1. Recursion Depth Limit

• MAXIMUM 1 level deep: Parent Research Agent → Sub-Agent(s)
• Sub-Agents CANNOT spawn further research agents
• Any research task requiring more depth must be handled by the parent

2. Sub-Agent Tier Restrictions

When a Research Agent spawns Sub-Agents:

• ✅ ALLOWED: QuickResearch (1 agent, FREE)
• ✅ ALLOWED: StandardResearch (3 agents, ~$0.01)
• ❌ FORBIDDEN: ExtensiveResearch in Sub-Agents

3. ExtensiveResearch Requires Human Approval

• ExtensiveResearch (~$0.10-0.50) MUST get explicit user confirmation before executing
• Sub-Agents CANNOT auto-trigger expensive research modes
• Pattern: Ask user → Get confirmation → Then spawn

4. Enforcement

When implementing research workflows:

// Parent agent (you) decides tier
IF cost < $0.01 AND no_sub_agents_needed:
  → Use QuickResearch or StandardResearch
  
IF cost > $0.01 OR complex multi-angle:
  → Ask user for ExtensiveResearch approval
  → Only spawn Sub-Agents with Quick/Standard modes
  → Sub-Agents return results to parent
  → Parent synthesizes (no further spawning)

Violation: If you see a plan with 3+ levels of research agents, STOP and restructure.