agentic-engineering-agency/specsafe-explore

Pre-specification exploration mode - research and spike before creating formal spec.

When to use:

• Unclear requirements or scope
• Need to research technical approach
• Want to prototype before committing
• Evaluating feasibility of a feature
• Comparing alternative solutions

Input: Feature idea or problem statement

Steps

1. Understand the problem

   Clarify with user:

   • What problem are we solving?
   • Who are the stakeholders?
   • What are the constraints?
   • What's the expected timeline?

2. Research existing solutions

   Explore:

   • Industry best practices
   • Similar features in other products
   • Open source libraries available
   • Design patterns applicable

3. Spike/prototype (optional)

   If technical feasibility is uncertain:

   • Create minimal proof-of-concept
   • Test critical assumptions
   • Measure performance characteristics
   • Document findings

   Keep spike code separate:

   spikes/
   └── <feature-name>-exploration/
       ├── README.md
       └── [prototype files]
   

4. Document findings

   Create exploration notes:

   # Exploration: [Feature Name]
   
   ## Problem Statement
   [Clear description of the problem]
   
   ## Research Findings
   - [Finding 1]
   - [Finding 2]
   
   ## Options Considered
   
   ### Option A: [Name]
   - Pros: ...
   - Cons: ...
   - Effort: [Estimate]
   
   ### Option B: [Name]
   - Pros: ...
   - Cons: ...
   - Effort: [Estimate]
   
   ## Recommendation
   [Recommended approach with justification]
   
   ## Open Questions
   - [Question 1]
   - [Question 2]
   
   ## Next Steps
   - [ ] Get stakeholder feedback
   - [ ] Create formal spec
   - [ ] Break down into smaller specs
   

5. Evaluate feasibility

   Assess:

   • Technical complexity
   • Resource requirements
   • Timeline feasibility
   • Risk factors
   • Dependencies

6. Recommend path forward

   Decide:

   • ✅ Create spec: Ready to formalize
   • 🔄 More exploration: Need additional research
   • ❌ Not viable: Too complex or not aligned
   • ✂️ Split: Break into multiple smaller specs

7. Transition to formal spec (if ready)

   If exploration yields clear requirements:

   Exploration complete → /specsafe:new <feature-name>
   

Output

After exploration:

• 📋 Exploration notes documented
• 🔍 Research findings recorded
• 💡 Options compared with pros/cons
• 🎯 Feasibility assessment
• 📊 Effort estimates
• 🚦 Recommendation (proceed/pivot/abandon)

If proceeding:

• 📋 Prompt: "Ready to create spec? Run /specsafe:new <feature-name>"

If more exploration needed:

• 📋 Prompt: "Continue exploration or schedule spike?"

Guardrails

• Exploration is time-boxed (suggest 2-4 hours max)
• Document everything for future reference
• Don't over-engineer prototypes (throwaway code OK)
• Be honest about risks and unknowns
• Involve stakeholders before committing to spec
• Exploration ≠ Implementation (no production code)

Example

User: /specsafe:explore "real-time collaboration"
→ Problem: Users need simultaneous editing
→ Research: Operational Transform vs CRDT
→ Spike: Prototype OT implementation
→ Findings: CRDT better for our use case
→ Options:
→   A: Use Yjs library (3 days)
→   B: Build custom CRDT (2 weeks)
→   C: Use Liveblocks SaaS (1 day + ongoing cost)
→ Recommendation: Option A (Yjs)
→ 
→ 📋 Ready to spec? /specsafe:new real-time-collaboration

vs Formal Spec

| Exploration | Formal Spec | |-------------|-------------| | Unclear requirements | Clear requirements | | Research needed | Research done | | May be discarded | Committed to implement | | Spike code OK | Production code only | | Time-boxed | Fully estimated | | Optional artifact | Required artifact |