planning-research

name	planning-research
description	Conducts market, domain, and technical research to validate assumptions and fill knowledge gaps before brief or PRD creation. Uses Perplexity deep research (perplexity_research) for web research, Context7 for library/framework documentation, and Tavily for cross-validation. Can run domain, market, or technical research independently or combined.

Planning Research

When This Skill Runs

• After brainstorming, to validate the selected direction
• Before brief creation, to ground the brief in evidence
• Before PRD creation, to validate technical feasibility
• When the human says "research X before we proceed"

Research Types

Domain Research

• Industry landscape and trends
• Regulatory requirements and compliance
• Domain terminology and standards
• Existing solutions and their limitations

Market Research

• Target audience validation (do they exist? how many?)
• Customer pain point verification (are these real problems?)
• Competitive analysis (what exists? what's missing?)
• Pricing and business model patterns

Technical Research

• Technology stack evaluation (what's mature? what's risky?)
• Integration patterns (how do these components connect?)
• Architecture options (what patterns fit this problem?)
• Library/framework selection (versions, trade-offs, community health)

Research Process

1. Receive research brief -- what questions need answering
2. Spawn research-agent with Perplexity (deep research via perplexity_research) + Context7 + Tavily MCP tools
3. Follow AGENTS.md query construction rules (explicit search directives, parent org context, alternative terms, source suggestions)
4. Cross-reference findings across at least 2 independent sources
5. Date-stamp all findings -- technology landscapes change rapidly
6. Write research report to .factory/planning/research-report.md
7. Flag uncertainties -- what couldn't be verified, what needs human judgment

Quality Gate

• Research report produced with structured findings per research type
• All findings cite specific sources with URLs and dates
• Inconclusive areas and uncertainties explicitly flagged
• Cross-referenced across at least 2 independent sources where possible

Failure Modes

• If MCP tools (Perplexity, Context7, Tavily) are unavailable: use training data with explicit "UNVERIFIED -- based on training data, not live research" disclaimer
• If research yields no relevant results for a research type: document the gap and recommend alternative research approaches
• If sources contradict each other: present both sides with evidence strength assessment

Output Artifacts

• .factory/planning/research-report.md
• .factory/planning/research-sources.md (citations and links)