Deep Research Mission Workflow

Conduct systematic research investigations using Spec Kitty's research mission template.

When to Use Research Mission

  • Investigating technical approaches before implementation
  • Literature reviews and technology comparisons
  • Evidence-based decision making for architecture choices
  • Academic or industry research projects
  • Due diligence on tools, frameworks, or patterns

Switching to Research Mode

cd my-research-project
spec-kitty mission list        # Show available missions
spec-kitty mission switch research  # Activate Deep Research Kitty

What changes:

  • Templates optimized for research workflows
  • Different artifact expectations (findings.md vs plan.md)
  • Research-focused command prompts
  • Evidence collection emphasis

Complete Research Workflow

1. Initialize Project (One-time)

spec-kitty init auth-research --mission research --ai claude
cd auth-research
claude

2. Define Research Question

/spec-kitty.specify

Investigate the optimal authentication patterns for serverless
applications with focus on:
- Token management strategies
- Session handling approaches
- Security considerations
- Performance implications

Compare solutions for AWS Lambda, Vercel Edge, and Cloudflare Workers.

Result: Creates kitty-specs/001-serverless-auth-study/spec.md with research objectives

3. Research Plan

/spec-kitty.plan

Survey these information sources:
- Academic papers on stateless authentication
- AWS, Vercel, Cloudflare documentation
- Open-source implementations (Auth0, Supabase)
- Security guidelines (OWASP, NIST)
- Performance benchmarks from industry blogs

Focus areas:
- JWT vs session tokens
- Token rotation strategies
- Cold start implications
- Multi-region session storage

Result: Creates research methodology in plan.md

4. Evidence Collection Phase

/spec-kitty.research

Result: Creates Phase 0 research artifacts:

kitty-specs/001-serverless-auth-study/
├── spec.md                      # Research objectives
├── plan.md                      # Methodology
├── research.md                  # Findings and analysis
├── data-model.md                # Key concepts and relationships
└── research/
    ├── evidence-log.csv         # Source tracking
    ├── comparison-matrix.md     # Side-by-side comparisons
    └── synthesis-notes.md       # Integration insights

5. Evidence Log Format

The research mission generates evidence-log.csv for tracking sources:

timestamp,source_type,citation,key_finding,confidence,notes
2025-01-15T10:30:00Z,paper,"Smith et al 2024, JWT Security",Token rotation reduces breach window,high,Peer-reviewed
2025-01-15T11:00:00Z,docs,"AWS Lambda Auth Docs",Sessions require external store,high,Official docs
2025-01-15T14:20:00Z,blog,"Auth0 Blog: Serverless Auth",Cold starts impact auth latency,medium,Industry observation

6. Generate Research Tasks

/spec-kitty.tasks

Result: Creates work packages for:

  • Literature review (by topic area)
  • Implementation analysis (by platform)
  • Comparison matrices (by criteria)
  • Synthesis and recommendations

Example tasks:

## WP01: AWS Lambda Authentication Patterns

### Subtasks
- [ ] T001: Review AWS Cognito integration patterns
- [ ] T002: Analyze custom JWT validation approaches
- [ ] T003: Document cold start mitigation strategies
- [ ] T004: Benchmark token validation performance

## WP02: Cross-Platform Comparison Matrix

### Subtasks
- [ ] T005: Compare token storage options (Lambda vs Edge)
- [ ] T006: Evaluate session management tradeoffs
- [ ] T007: Document security model differences
- [ ] T008: Create recommendation framework

7. Execute Research

/spec-kitty.implement

Research implementation workflow:

  1. Moves work package to "doing"
  2. Agent conducts research, documents findings in research.md
  3. Updates evidence-log.csv with sources
  4. Creates comparison matrices as needed
  5. Moves to "for_review" when complete

8. Synthesize Findings

/spec-kitty.review

Review research outputs for:

  • Evidence quality and citation accuracy
  • Comparison fairness and completeness
  • Logical flow of arguments
  • Actionable recommendations

9. Finalize Research

/spec-kitty.accept

Validates:

  • All evidence logged with sources
  • Comparison matrices complete
  • Recommendations backed by evidence
  • Findings reproducible from evidence log

Research Artifacts Explained

spec.md (Research Objectives)

  • Research questions
  • Hypothesis (if applicable)
  • Success criteria for research
  • Scope boundaries

plan.md (Methodology)

  • Information sources to consult
  • Analysis framework
  • Comparison criteria
  • Quality standards for evidence

research.md (Findings)

  • Key discoveries organized by theme
  • Evidence synthesis
  • Comparison results
  • Recommendations with rationale

data-model.md (Concepts)

  • Key terms and definitions
  • Relationships between concepts
  • Mental models and frameworks
  • Taxonomies and categorizations

evidence-log.csv (Sources)

  • Timestamp of collection
  • Source type (paper, docs, blog, etc.)
  • Full citation
  • Key finding extracted
  • Confidence level (high/medium/low)
  • Additional notes

Switching Back to Development

After research completes:

# Accept research findings
/spec-kitty.accept
/spec-kitty.merge

# Switch back to software development mode
spec-kitty mission switch software-dev

# Start implementation based on research
/spec-kitty.specify
Implement JWT-based authentication for serverless API...

Example: Technology Evaluation

Research Question: Which database is best for our use case?

Workflow:

# 1. Define question
/spec-kitty.specify
Compare PostgreSQL, MongoDB, and DynamoDB for:
- Read-heavy workload (10:1 read:write ratio)
- JSON document storage
- <100ms query latency requirement
- Cost at 10M requests/month

# 2. Methodology
/spec-kitty.plan
Evaluate using:
- Official benchmarks
- Case studies from similar scale
- Pricing calculators
- Community discussions

# 3. Collect evidence
/spec-kitty.research

# 4. Generate tasks
/spec-kitty.tasks
Creates: WP01 (PostgreSQL analysis), WP02 (MongoDB analysis),
         WP03 (DynamoDB analysis), WP04 (Comparison matrix)

# 5. Execute research
/spec-kitty.implement (repeat for each WP)

# 6. Result: evidence-backed database recommendation

Benefits of Research Mission

  1. Systematic Evidence Collection

    • No missed sources
    • Auditable research trail
    • Reproducible findings

  2. Quality Control

    • Evidence confidence ratings
    • Peer review via /spec-kitty.review
    • Citation requirements

  3. Decision Documentation

    • Future teams understand why decisions were made
    • Research reusable for similar questions
    • Recommendations traceable to evidence

  4. Parallel Research

    • Multiple agents can research different aspects
    • Dashboard shows research progress
    • Work packages prevent duplication

Tips for Research Mission

  • Start specific: Narrow research questions get better results
  • Log as you go: Update evidence-log.csv during research, not after
  • Use confidence levels: Distinguish strong evidence from speculation
  • Create matrices: Side-by-side comparisons force thorough analysis
  • Synthesize early: Don't collect forever - analyze iteratively
  • Switch missions: Research → Development → Research as needed

Common Research Patterns

Technology Selection:

  • WP01-WP0N: One work package per option
  • WP Last: Comparison matrix and recommendation

Literature Review:

  • WP01: Search and source collection
  • WP02-WP0N: Analysis by theme/topic
  • WP Last: Synthesis and gaps analysis

Best Practices Study:

  • WP01: Industry standards research
  • WP02: Case studies collection
  • WP03: Pattern extraction
  • WP04: Recommendations for context

Exiting Research Mission

# View current mission
spec-kitty mission current

# List available missions
spec-kitty mission list

# Switch back to development
spec-kitty mission switch software-dev

# Verify switch
spec-kitty mission current  # Should show "Software Dev Kitty"

Note: Switching missions changes command behaviors and template expectations. Plan accordingly.

Edit this page