| name | research |
| description | Investigates patterns, gathers context, and identifies best practices for workspace design. Use when starting a new workspace, researching workflow patterns, or before architecture planning. |
| triggers | ["research workflow","gather context","identify patterns","best practices"] |
Overview
Gather context and identify patterns before building. Research ensures the workspace design is informed by real requirements, not assumptions.
When to Use
- Phase 1 of the hybrid flow (always first)
- Before architecture planning
- When the user asks for a novel workflow type
- When existing patterns do not fit the use case
When Not to Use
- After architecture is already planned (use architecture sub-skill)
- When workspace structure already exists (use validation sub-skill)
- For simple file creation (use direct file operations)
The Iron Law
NO BUILD WITHOUT RESEARCH
NO GENERIC FINDINGS
NO SKIPPING INPUT/OUTPUT ANALYSIS
NO ASSUMPTIONS WITHOUT EVIDENCE
Scope Guardrails
- Research the workflow needed to produce outcomes, not the product implementation stack.
- Convert domain asks (for example, "football predictor") into stageable workflow responsibilities.
- Keep research outputs markdown-first and suitable for numbered folder CONTEXT contracts.
- Do not produce backend/frontend/data-model implementation plans in this phase.
The Process
- Identify workflow type - Determine what process is being automated as a file-structured markdown workflow.
- Research similar patterns - Review existing workspaces, docs, and best practices.
- Identify key stages - Define the natural workflow phases.
- Determine inputs and outputs - Capture what goes in and what markdown artifacts come out at each stage.
- Identify tooling needs - List tools commonly needed for this workflow.
- Document findings - Create a concise research summary for architecture.
Red Flags
- Research is too generic and not tied to the requested workflow
- Input and output analysis is missing
- Tooling assessment is missing
- Architecture starts before research findings are complete
- Findings drift into app architecture, model design, or runtime repository scaffolding
Anti-Rationalization Table
| Thought | Reality |
|---|
| "I already know this workflow type" | Knowledge is not research. Document findings for the next agent. |
| "Research is taking too long" | Research prevents wasted build time. Be thorough. |
| "I will figure it out while building" | Building without research produces generic, non-optimal workspaces. |
| "The user will clarify later" | Ask now. Ambiguous requirements produce ambiguous workspaces. |
Sub-Skill Dispatch
status = passed -> nextSkill = architecture.
status = failed (research incomplete but recoverable) -> nextSkill = none and request missing inputs before rerun.
status = escalated (blocking ambiguity or conflicting constraints) -> nextSkill = none.
Report Format
{
"skill": "research",
"status": "passed",
"timestamp": "2026-04-08T00:00:00Z",
"findings": ["Identified three reusable workflow stage patterns"],
"recommendations": ["Use a three-stage layout with explicit input/output boundaries"],
"metrics": {
"patternsIdentified": 3,
"stagesIdentified": 3
},
"nextSkill": "architecture"
}
Allowed status values: passed, failed, escalated.
Allowed nextSkill values: architecture, none.
Integration
- Feeds architecture with concrete findings and stage proposals.
- Reduces rework by grounding structure decisions in evidence.