| name | research |
| description | Competitive analysis, technical research, and market validation for new features |
| version | 1.0.0 |
| author | potato |
| triggers | {"patterns":["research","竞品","调研","market research","competitive analysis"]} |
| tags | ["research","analysis","strategy"] |
| priority | 60 |
| always_load | false |
| max_body_size | 4096 |
SKILL: Research Phase
Conduct competitive analysis, technical research, and market validation before drafting requirements.
Trigger Conditions
This skill activates when:
- Starting a new feature that needs research
- User asks for "research", "竞品", "调研", "market research", or "competitive analysis"
- GID ritual reaches the research phase
Input
- Feature name/description from capture-idea phase
- Any existing context from
.gid/features/{feature}/idea.md
Output
Create docs/RESEARCH-{feature-name}.md with the following structure:
# Research: {feature-name}
## Executive Summary
2-3 sentence overview of findings and recommendation.
## Competitive Landscape
### {Competitor 1}
- **URL**:
- **What it does**:
- **Key features**:
- **Strengths**:
- **Weaknesses**:
- **Pricing**: (if applicable)
- **User sentiment**: (from reviews, forums, etc.)
### {Competitor 2}
...
### Competitive Matrix
| Feature | Us (planned) | Competitor 1 | Competitor 2 |
|---------|--------------|--------------|--------------|
| ... | ... | ... | ... |
## Technical Approaches
### Option A: {approach name}
- **Description**:
- **Pros**:
- **Cons**:
- **Effort estimate**:
- **Key libraries/dependencies**:
### Option B: {approach name}
...
### Technical Recommendation
Which approach to use and why.
## Market Signals
### User Demand Evidence
- Forum discussions, Reddit threads, HN comments
- GitHub issues/stars on related projects
- Search volume trends (if available)
### Pain Points Found
- What problems users are reporting
- What's missing in existing solutions
### Market Size Indicators
- Target audience size estimates
- Growth trends in the space
## Prior Art
### Open Source Projects
- Relevant repos with brief assessment
### Academic Papers / Blog Posts
- Key technical resources that inform the design
### Lessons Learned from Others
- What worked/failed for similar projects
## Key Insights
1. **Insight 1**: [actionable takeaway]
2. **Insight 2**: [actionable takeaway]
3. **Insight 3**: [actionable takeaway]
4. **Insight 4**: [actionable takeaway]
5. **Insight 5**: [actionable takeaway]
## Recommendation
### Go / No-Go Assessment
**Recommendation**: [GO / NO-GO / PIVOT]
**Reasoning**:
- [Key factor 1]
- [Key factor 2]
- [Key factor 3]
### If GO, Key Success Factors
- What must be true for this to succeed
- What differentiates us from competition
### If NO-GO, Alternatives
- What should we do instead
Research Process
Step 0: Codebase Discovery (MANDATORY before any research)
→ find . -type f \( -name "*.rs" -o -name "*.ts" -o -name "*.py" \) | xargs grep -li "{feature_keywords}" 2>/dev/null
→ find . -path "*/packages/*" -o -path "*/crates/*" -o -path "*/src/*" | grep -i "{feature_name}"
→ Check if implementation already exists under a different name/path
→ If files found:
- Read them to understand what's already built
- Document in research output: "## Existing Implementation" section
- Assess: is this task already done? partially done? needs different approach?
- If fully implemented: STOP research, report "already exists at {path}"
→ If nothing found: proceed to Step 1
Why this step exists: We once reimplemented a TypeScript MCP server in Rust because we didn't check packages/mcp/ first. This step is free (no LLM, just grep) and prevents wasted work.
Step 1: Competitive Analysis
→ web_search("{feature} alternatives")
→ web_search("{feature} competitors")
→ web_search("{feature} open source")
→ For each competitor found:
→ web_fetch(competitor_url) to understand features
→ web_search("{competitor} reviews") for user sentiment
Step 2: Technical Research
→ web_search("{feature} implementation")
→ web_search("{feature} architecture")
→ web_search("{feature} rust library" or "{feature} python library")
→ Check GitHub trending/search for relevant projects
→ Evaluate feasibility based on existing tools/libraries
Step 3: Market Validation
→ web_search("{feature} demand")
→ web_search("{feature} reddit" or "{feature} hacker news")
→ web_search("{feature} problems" or "{feature} pain points")
→ Look for patterns in user complaints/requests
Step 4: Prior Art
→ web_search("{feature} paper" or "{feature} research")
→ web_search("{feature} blog post" or "{feature} tutorial")
→ GitHub search for related projects, check their issues/discussions
Step 5: Synthesize & Recommend
→ Compile findings into structured document
→ Identify patterns across sources
→ Make go/no-go recommendation with clear reasoning
→ If GO, identify key differentiators and success factors
Notes
- Speed over perfection — This is research to inform decisions, not an exhaustive academic survey
- Focus on actionable insights — Every section should help make better product decisions
- Be honest about unknowns — If data is scarce, say so rather than speculating
- Quantify when possible — GitHub stars, forum upvotes, search trends > vague assertions
- Document sources — Include URLs so findings can be verified
- Time-box — Aim for 20-30 minutes of research, not hours
- Use 中英混用 naturally if researching Chinese market/competitors