| name | moai-domain-research |
| description | Market and ecosystem research specialist for /moai brain Phase 3. Executes parallel WebSearch + Context7 queries, handles tool failures gracefully, and produces structured research.md artifacts with cited sources and research limitations.
|
| when_to_use | Use for /moai brain Phase 3 market and ecosystem research: parallel WebSearch and Context7 queries, competitive-landscape analysis, source citation, and structured research.md artifacts with graceful tool-failure handling.
|
| license | Apache-2.0 |
| compatibility | Designed for Claude Code |
| allowed-tools | Read, Write, Edit, Grep, Glob, WebSearch, WebFetch, mcp__context7__resolve-library-id, mcp__context7__get-library-docs |
| user-invocable | false |
| metadata | {"version":"1.0.0","category":"domain","status":"active","updated":"2026-05-04","modularized":"false","tags":"research, web-search, context7, market-analysis, parallel-tools, brain","related-skills":"moai-domain-ideation, moai-foundation-thinking"} |
| progressive_disclosure | {"enabled":true,"level1_tokens":100,"level2_tokens":5000} |
Research Domain Specialist
Parallel research executor for the brain workflow's Phase 3. Issues WebSearch and Context7 tool calls simultaneously (single-message parallel call pattern per Anthropic best practice), handles partial failures gracefully, and assembles a structured research.md artifact.
Quick Reference
Core responsibilities:
- Execute WebSearch + Context7 in parallel (single message, multiple tool calls)
- Handle tool failures gracefully (partial-result tolerance)
- Produce
research.md with cited sources and explicit Research Limitations section
- Stay language/technology neutral
Key guarantees:
- [HARD] Tool calls are issued in parallel (single Claude message), not sequentially
- [HARD] Research failure does NOT abort Phase 3 — partial results are acceptable
- [HARD] Every source in research.md has a citation (URL or tool reference)
- [HARD] Research Limitations section present when any tool call fails or returns empty
Phase 3: Research Execution
Input
- Clarity-scored idea with user context from Phase 1
- Diverged concept map from Phase 2 (in-memory)
- Optional: existing
.moai/project/tech.md for tech-stack context (read-only)
Step 1: Query Design
Before issuing tool calls, design 2-4 targeted queries for each tool type:
WebSearch query design principles:
- Search for: existing solutions, market size, user pain points, competitive landscape
- Include both broad queries ("habit tracking apps market") and targeted queries ("habit tracking for seniors accessibility challenges")
- One query per major angle from Phase 2 diverge (use top 3 angles)
- Avoid technology-specific queries unless the user explicitly constrained to a tech stack
Context7 query design principles:
- Search for: relevant libraries, frameworks, or platforms in the solution space
- Focus on ECOSYSTEM tools (not specific language implementations) — e.g., query "habit tracking SDK" not "habit tracking React library"
- Use
resolve-library-id first, then get-library-docs for top matches
Step 2: Parallel Tool Calls
[HARD] Issue ALL prepared tool calls in a SINGLE Claude message. This is the parallel tool call pattern documented in Anthropic's tool-use documentation.
Pattern (pseudocode — actual tool syntax per Claude Code):
