ralph-prompt-research（拉尔夫提示研究）

Ralph Prompt Generator: Research & Analysis

Overview

Generates structured prompts for research, analysis, and planning tasks using the Ralph Wiggum technique. These prompts define clear research methodology, required sources, synthesis requirements, and concrete deliverables rather than code.

Best For:

• Codebase analysis and audits
• Security vulnerability assessments
• Technology research and comparison
• Migration planning
• Architecture design documents
• Performance analysis
• Dependency audits
• Documentation generation from code

Ralph Philosophy: This generator embraces the principle that failures are deterministic and fixable. Each research iteration builds on previous findings. Don't fear incomplete first passes—they're expected and provide data for deeper analysis in subsequent iterations.

Quick Start

Input Required:

1. Research objective (what you need to understand/analyze)
2. Scope (what to include/exclude)
3. Deliverable format (report, plan, document)
4. Completion promise

Generate prompt with:

Generate a Ralph research prompt for:
Objective: [What to research/analyze]
Scope: [What's included]
Deliverable: [What to produce]
Promise: [COMPLETION_PHRASE]

Prompt Generation Workflow

Step 1: Define Research Scope

Research Definition Template:

OBJECTIVE: [What question to answer or what to analyze]
SCOPE:
  Include: [What's in scope]
  Exclude: [What's explicitly out of scope]
SOURCES: [Where to look for information]
DELIVERABLE: [What artifact to produce]
SUCCESS: [How to know research is complete]

Step 2: Map Research Phases

Standard research phases:

Step 3: Define Completion Criteria

Research completion criteria must be:

• Based on coverage (examined all relevant items)
• Based on depth (sufficient analysis per item)
• Based on deliverable completeness (all sections written)

Examples:

Step 4: Structure the Prompt

Use this template:

# Research: [Research Title]

## Objective
[Clear statement of what this research aims to answer or analyze]

## Scope

### In Scope
- [Item 1 to analyze]
- [Item 2 to analyze]
- [Item 3 to analyze]

### Out of Scope
- [Explicitly excluded 1]
- [Explicitly excluded 2]

## Sources to Examine
- [Source type 1]: [Where to find]
- [Source type 2]: [Where to find]
- [Source type 3]: [Where to find]

---

## Phase 1: Discovery

### Objective
Create comprehensive inventory of [what to inventory].

### Tasks
1. [Discovery task 1]
2. [Discovery task 2]
3. [Discovery task 3]

### Deliverable: Inventory
Create `[filename].md` with:

[Inventory Title]

Items Found

### Phase 1 Success Criteria

- All [scope items] examined
- Inventory file created
- No areas unexplored

### Phase 1 Checkpoint

Document in inventory file:

DISCOVERY COMPLETE:

• Items found: [count]
• Areas examined: [list]
• Coverage: Complete/Partial

---

## Phase 2: Analysis

### Objective

Deep analysis of each discovered item.

### Analysis Framework

For each item, analyze:

- [Dimension 1]: [What to examine]
- [Dimension 2]: [What to examine]
- [Dimension 3]: [What to examine]

### Tasks

1. Analyze each item against framework
2. Document findings per item
3. Identify patterns across items
4. Note anomalies or concerns

### Deliverable: Analysis Notes

Create `[analysis-filename].md` with:

Analysis: [Item Name]

[Dimension 1]

[Findings]

[Dimension 2]

[Findings]

[Dimension 3]

[Findings]

Summary

[Key takeaways for this item]

### Phase 2 Success Criteria

- All inventory items analyzed
- Analysis notes for each item
- Patterns identified
- Anomalies documented

### Phase 2 Checkpoint

ANALYSIS COMPLETE:

• Items analyzed: [X/Y]
• Key patterns: [list]
• Concerns identified: [count]

---

## Phase 3: Synthesis

### Objective

Synthesize findings into coherent conclusions.

### Tasks

1. Review all analysis notes
2. Identify overarching themes
3. Draw conclusions from patterns
4. Prioritize findings by impact
5. Formulate key insights

### Deliverable: Findings Summary

Create `[findings-filename].md` with:

Research Findings: [Topic]

Executive Summary

[2-3 paragraphs summarizing key findings]

Key Findings

1. [Finding 1]: [Description and evidence]
2. [Finding 2]: [Description and evidence]
3. [Finding 3]: [Description and evidence]

Patterns Observed

• [Pattern 1]
• [Pattern 2]

Areas of Concern

1. [Concern 1]: [Details]
2. [Concern 2]: [Details]

Data Summary

• Total items examined: [X]
• Categories: [list]
• Analysis coverage: [X%]

### Phase 3 Success Criteria

- All findings synthesized
- Conclusions supported by evidence
- Priorities established
- Summary document complete

---

## Phase 4: Recommendations

### Objective

Provide actionable recommendations based on findings.

### Tasks

1. Develop recommendations for each finding
2. Prioritize by impact and effort
3. Identify dependencies between recommendations
4. Create action timeline (if applicable)

### Deliverable: Recommendations

Add to findings document or create separate:

Recommendations

High Priority (Address Immediately)

1. [Recommendation]

- Rationale: [Why] - Effort: [Low/Medium/High] - Impact: [Low/Medium/High] - Dependencies: [Any]

Medium Priority (Plan for Soon)

[...]

Low Priority (Future Consideration)

[...]

Implementation Order

1. First: [Recommendation X]
2. Then: [Recommendation Y]
3. Finally: [Recommendation Z]

### Phase 4 Success Criteria

- All findings have recommendations
- Priorities assigned
- Implementation order defined
- Dependencies identified

---

## Phase 5: Final Documentation

### Objective

Produce final polished deliverable.

### Tasks

1. Compile all sections into final document
2. Add executive summary
3. Review for completeness
4. Ensure actionability
5. Format consistently

### Deliverable: Final Report

Create `[final-report].md` with all sections:

1. Executive Summary
2. Scope and Methodology
3. Findings
4. Analysis
5. Recommendations
6. Appendix (raw data if needed)

### Phase 5 Success Criteria

- All sections complete
- Executive summary written
- Consistent formatting
- No TODO items
- Actionable and clear

---

## Final Verification

### Completeness Checklist

- Phase 1: Discovery complete, inventory exists
- Phase 2: All items analyzed, notes exist
- Phase 3: Findings synthesized, summary exists
- Phase 4: Recommendations provided, prioritized
- Phase 5: Final document complete

### Quality Checklist

- All conclusions supported by evidence
- Recommendations are actionable
- Document is well-organized
- No sections are missing content
- Executive summary captures key points

## Completion

When ALL of the following are true:

- All 5 phases completed
- All deliverables created
- Completeness checklist verified
- Quality checklist verified

Output: [COMPLETION_PHRASE]

## Progress Tracking

RESEARCH PROGRESS: Phase 1 (Discovery): [COMPLETE/IN PROGRESS] Phase 2 (Analysis): [COMPLETE/IN PROGRESS/NOT STARTED] Phase 3 (Synthesis): [COMPLETE/IN PROGRESS/NOT STARTED] Phase 4 (Recommendations): [COMPLETE/IN PROGRESS/NOT STARTED] Phase 5 (Documentation): [COMPLETE/IN PROGRESS/NOT STARTED]

Items examined: [X/Y] Deliverables created: [list]

## If Stuck

- Document what information is missing
- List sources already examined
- Identify the specific gap
- Do NOT skip analysis or output false promise

Example Prompts

Example 1: Security Audit

Input:

• Objective: Identify security vulnerabilities in authentication module
• Scope: src/auth/ directory
• Deliverable: Security audit report with findings and fixes
• Promise: SECURITY_AUDIT_COMPLETE

Generated Prompt:

# Research: Authentication Security Audit

## Objective
Identify security vulnerabilities in the authentication module following OWASP guidelines and provide remediation recommendations.

## Scope

### In Scope
- All files in src/auth/
- Authentication flows (login, signup, password reset)
- Session management
- Token handling
- Input validation

### Out of Scope
- Authorization (permissions, roles)
- Third-party auth providers
- Frontend security

## Sources to Examine
- Code files: src/auth/**/*.ts
- Tests: tests/auth/**/*.ts
- Configuration: auth-related config files
- Dependencies: auth-related npm packages

---

## Phase 1: Discovery

### Tasks
1. List all files in auth module
2. Identify all entry points (routes)
3. Map authentication flows
4. Identify external dependencies
5. List all sensitive operations

### Deliverable: auth-inventory.md

Auth Module Inventory

Files

Dependencies

| Package | Version | Security Notes |

Authentication Flows

1. Login: [path]
2. Signup: [path]

[...]

### Phase 1 Success

- All auth files listed
- All entry points identified
- Flows mapped
- Dependencies inventoried

---

## Phase 2: Analysis

### Security Analysis Framework

For each component, check:

- **A1 Injection**: SQL, NoSQL, command injection risks
- **A2 Broken Auth**: Weak passwords, session issues
- **A3 Sensitive Data**: Password storage, token exposure
- **A4 XXE**: XML parsing vulnerabilities
- **A5 Access Control**: Proper authorization checks
- **A6 Misconfiguration**: Insecure defaults
- **A7 XSS**: User input handling
- **A8 Deserialization**: Unsafe object handling
- **A9 Components**: Vulnerable dependencies
- **A10 Logging**: Insufficient monitoring

### Tasks

1. Analyze each file against OWASP framework
2. Check password hashing implementation
3. Review JWT token handling
4. Examine session management
5. Check input validation

### Deliverable: security-analysis.md

Per-file analysis with vulnerabilities found.

---

## Phase 3: Synthesis

### Tasks

1. Categorize all vulnerabilities by severity
2. Identify systemic issues
3. Calculate risk scores
4. Prioritize by exploitability

### Deliverable: security-findings.md

Security Findings

Critical (Requires Immediate Action)

• [Vulnerability]: [Location] - [Impact]

High

[...]

Medium

[...]

Low

[...]

Summary Statistics

• Critical: X
• High: X
• Medium: X
• Low: X

---

## Phase 4: Recommendations

### Tasks

1. Provide fix for each vulnerability
2. Prioritize fixes by risk
3. Estimate effort for each fix
4. Identify quick wins

### Deliverable: remediation-plan.md

Remediation Plan

Immediate Actions

1. [Fix]: [How] - [Effort]

Short-term (This Sprint)

[...]

Medium-term

[...]

---

## Phase 5: Final Report

### Deliverable: AUTH-SECURITY-AUDIT.md

Complete report with:

1. Executive Summary
2. Methodology
3. Findings by Severity
4. Detailed Vulnerability Descriptions
5. Remediation Plan
6. Verification Steps

---

## Completion

When audit report complete with all vulnerabilities documented and remediation provided:

Output: SECURITY_AUDIT_COMPLETE

Example 2: Technology Comparison

Input:

• Objective: Compare React vs Vue vs Svelte for new frontend
• Scope: Team skills, project requirements, ecosystem
• Deliverable: Comparison report with recommendation
• Promise: FRAMEWORK_COMPARISON_COMPLETE

Generated Prompt:

# Research: Frontend Framework Comparison

## Objective
Compare React, Vue, and Svelte to recommend the best framework for the new frontend project based on team skills, project requirements, and ecosystem.

## Scope

### In Scope
- Learning curve for team
- Performance characteristics
- Ecosystem (libraries, tools)
- Community and support
- Project requirements fit
- Long-term maintainability

### Out of Scope
- Mobile development (React Native, etc.)
- SSR frameworks (Next.js, Nuxt, SvelteKit)
- Backend considerations

---

## Phase 1: Discovery

### Tasks
1. Document team's current skills
2. List project requirements
3. Identify key evaluation criteria
4. Gather framework documentation

### Deliverable: evaluation-criteria.md

Evaluation Framework

Team Context

• Current experience: [frameworks]
• Available learning time: [duration]
• Team size: [N]

Project Requirements

• [Requirement 1]
• [Requirement 2]

[...]

Evaluation Criteria

[...]

---

## Phase 2: Analysis

### Analysis Framework

For each framework:

- **Learning Curve**: Time to productivity
- **Performance**: Bundle size, runtime speed, memory
- **Ecosystem**: Available libraries, tools, IDE support
- **Community**: Documentation, Stack Overflow, GitHub activity
- **Hiring**: Developer availability
- **Architecture**: How it fits project patterns

### Deliverable: framework-analysis.md

React Analysis

Learning Curve

[Analysis]

Performance

[Analysis]

[... repeat for Vue and Svelte]

---

## Phase 3: Synthesis

### Tasks

1. Score each framework on each criterion
2. Apply weights
3. Calculate overall scores
4. Identify dealbreakers

### Deliverable: comparison-matrix.md

Framework Comparison Matrix

[...] | Weighted Total | | X.X | X.X | X.X |

---

## Phase 4: Recommendations

### Deliverable: recommendation.md

Framework Recommendation

Recommendation: [Framework]

Rationale

[Why this framework is best for this team and project]

Advantages

• [Advantage 1]
• [Advantage 2]

Concerns and Mitigations

• [Concern 1]: [How to address]

Implementation Approach

1. [Step 1]
2. [Step 2]

[...]

---

## Phase 5: Final Report

### Deliverable: FRAMEWORK-COMPARISON-REPORT.md

---

## Completion

When complete comparison with recommendation:

Output: FRAMEWORK_COMPARISON_COMPLETE

Best Practices

Research Methodology

• Define scope explicitly (what's in/out)
• Use consistent analysis framework
• Document evidence for conclusions
• Separate findings from opinions

Deliverables

• Create artifacts as you go
• Use consistent format
• Include data to support conclusions
• Make recommendations actionable

DO:

• Define clear scope boundaries
• Create deliverables at each phase
• Support conclusions with evidence
• Provide actionable recommendations
• Track progress systematically

DON'T:

• Skip the discovery phase
• Make recommendations without evidence
• Leave sections incomplete
• Output promise before documentation

Integration with Ralph Loop

/ralph-wiggum:ralph-loop "[paste generated prompt]" --completion-promise "YOUR_PROMISE" --max-iterations 50

Recommended iterations by research complexity:

• Simple audit (single module): --max-iterations 30-40
• Comparison (3-5 options): --max-iterations 40-60
• Full codebase analysis: --max-iterations 70-100
• Complex planning: --max-iterations 60-80

For single-task prompts, see ralph-prompt-single-task. For multi-task prompts, see ralph-prompt-multi-task. For project-level prompts, see ralph-prompt-project.