Research Skill

research OVERVIEW actor: {{actor}} | mandate: Conduct systematic investigation and analysis | bounds: Research and analysis only

ORIENTATION Confirm .spec/constitution.md for current architecture and delivery guardrails. Review .spec/glossary.md entries relevant to research to keep terminology aligned. Load the {{actor}} persona brief from .spec/agents/{{actor}}.agent.md for format expectations.

LIFECYCLE TRANSITION from: ANY ➜ to: SAME ACCEPTS_FROM: [] tag: [research] (append to story ## Lifecycle Log)

WHEN TO RUN Research investigations needed for decision making Hypotheses need validation with quantitative/qualitative data {{research_domain}} patterns and constraints require analysis {{research_domain}} research or competitive analysis needed Best practices investigation required for implementation

REQUIRED INPUTS story_id research_type ({{research_types}} - includes product-discovery) research_questions (specific questions to answer) research_context (decision requirements or problems) {{research_domain_inputs}} MCP tooling: [WebSearch, WebFetch, Read]

CONTEXT PACK <<story.current_state>> <<personas.active_assignments>> <<recent_transition_log>> <<<GLOSSARY(term=research)>>>

EXECUTION ALGORITHM Validate prerequisites (research_type, research_questions, research_context). If anything missing → BLOCKED format below. Collect evidence based on research type: {{research_mode_evidence}} Execute analysis methodology appropriate to research type: {{research_mode_analysis}} Generate insights and recommendations based on findings. Document research methodology and evidence sources. Provide actionable recommendations for decision-making. Draft gate artifact using the structure in ARTIFACT OUTPUT. Append the TRANSITION LOG entry to the story's ## Lifecycle Log, matching persona tone. Update glossary/constitution if new terminology or workflows were introduced.

ARTIFACT OUTPUT === Research Summary === summary: inputs:research_type= research_questions= research_context= evidence:<analysis_type>|result=<validated/supported/identified>|ref=<path_to_research> risks:[ ]<risk_description>|owner=|mitigation= next_steps:<follow-up needed or n/a> === END Research Summary ===

TRANSITION LOG TEMPLATE [TRANSITION|research] by {{actor}} MODE: tolerant FROM_STATE: <current_state> TO_STATE: SAME WHY:

Research investigation required for informed decision making
{{research_type}} analysis needed to answer critical questions OUTPUT: === Research Summary === summary:Conducted {{research_type}} research and provided evidence-based recommendations. inputs:research_type={{research_type}} research_questions=docs/questions.md research_context=decision-context.pdf evidence:{{research_evidence_type}}|result=hypothesis_supported|ref=research/{{research_type}}-report-{{date}}.pdf risks:[ ]{{research_risk}}|owner={{actor}}|mitigation={{research_mitigation}} next_steps:Proceed with implementation using validated research findings. === END Research Summary === FOLLOW-UP:
Apply research findings - owner={{actor}} - due={{date}}

BLOCKED FORMAT BLOCKED(missing_inputs=[research_questions, research_context], unblock_steps=[define_questions, clarify_context])

GUARDRAILS Keep entries <=120 chars per line for CLI readability. All findings must be supported by evidence and appropriate citations. Research methodology must be clearly documented and reproducible. Limitations and assumptions must be explicitly stated. {{research_domain_guardrails}} Conflicting information must be acknowledged and resolved. Update .spec/glossary.md if you introduce new terms, channels, or artifacts.

Research Templates

Template Selection Mechanism

Agents can specify research type in two ways:

Skill Specification: Use research skill with research_type parameter
Intent Interpretation: System determines research type from context and questions

Usage Examples

bash

# Direct template specification
exec story=PROJECT-001 skill=research research_type=analytics

# System interpretation based on questions and context
exec story=PROJECT-001 skill=research  # Analyzes research_questions to determine type

Template Definitions

1. Analytics Research Template

Intended for: Any agent with data-driven questions When to use: Hypothesis validation, metrics analysis, quantitative insights Parameters:

research_type: analytics
actor: {{actor}} (flexible based on context)
research_types: analytics/technical/market/competitive
research_domain_inputs: data_sources (available data and analytics)
research_mode_evidence: | Analytics: data sources, analytics reports, metrics
research_mode_analysis: | Analytics: Statistical analysis, hypothesis testing, data validation
research_evidence_type: customer_analysis
research_risk: Data quality limitations
research_mitigation: additional_data_collection
research_domain_guardrails: Data sources must be clearly identified and validated.