Speckit Evaluate
Purpose: Evaluate the clarified spec against market research findings to optimize feature set before design phase.
When to Use
- •After Phase 3.5 (Research) is complete
- •Before Phase 4 (Design)
- •When research.md exists
Prerequisites
- • Spec exists at
.github/speckit/specs/{feature-name}.md - • Spec status is
clarified - • Research exists at
.github/speckit/specs/{feature-name}/research.md - • User has reviewed research findings
Execution
1. Delegate to Architect (Opus)
Use speckit-architect for deep reasoning evaluation:
code
task(
subagent_type="speckit-architect",
description="Evaluate spec against market research",
prompt="Evaluate this feature spec against the market research.
**Spec Path**: {spec path}
**Research Path**: {research.md path}
Read both files thoroughly, then provide evaluation:
## 1. Features to ADD
What's missing that competitors have or market expects?
- Feature name
- Rationale (why it's needed)
- Evidence (from research - which competitor, which trend)
- Impact: High/Medium/Low
- Effort estimate: Small/Medium/Large
## 2. Features to REMOVE
What's over-engineered, unnecessary, or doesn't align with market?
- Feature name
- Rationale (why remove)
- Evidence (competitors don't have it, not a best practice, etc.)
- Risk of keeping it
## 3. Features to MODIFY
What needs adjustment based on best practices?
- Feature name
- Current approach (from spec)
- Suggested change
- Rationale and evidence
- Impact: High/Medium/Low
## 4. Priority Assessment
Categorize ALL features (existing + suggested additions):
- **Must-Have (P0)**: Core value, competitors all have it
- **Should-Have (P1)**: Expected by users, differentiator
- **Nice-to-Have (P2)**: Enhancement, future consideration
## 5. Differentiation Strategy
Based on research, how should this feature stand out?
- Key differentiators to emphasize
- Unique value proposition
- Positioning against competitors
Be specific and cite evidence from research.md for each recommendation."
)
2. Present Evaluation to User
For each recommendation, ask user to Accept or Reject:
code
## Features to ADD
### 1. {Feature Name}
- **Rationale**: {why}
- **Evidence**: {from research}
- **Impact**: High | **Effort**: Medium
→ [Accept] / [Reject]
User decision: ___
### 2. {Feature Name}
...
## Features to REMOVE
...
## Features to MODIFY
...
3. Record Decisions
For each suggestion:
- •If Accept: Mark for inclusion in spec update
- •If Reject: Record user's reasoning
4. Update Spec
Apply accepted changes:
- •Add new features to spec
- •Remove rejected features
- •Modify adjusted features
- •Update priority order based on assessment
5. Save Evaluation
Save to: .github/speckit/specs/{feature-name}/evaluation.md
Include:
- •All recommendations
- •User decisions (accept/reject)
- •Reasoning for rejections
- •Final priority list
6. Finalize
- •Update spec status to
evaluated - •Confirm with user: "Spec is now finalized. Ready to proceed to Design?"
- •STOP and wait for explicit approval
Output Format
markdown
# Feature Evaluation: {Feature Name}
**Generated**: {date}
**Spec**: {spec path}
**Research**: {research path}
**Status**: Complete
## Evaluation Summary
| Category | Count | Accepted | Rejected |
|----------|-------|----------|----------|
| Add | 3 | 2 | 1 |
| Remove | 2 | 1 | 1 |
| Modify | 4 | 3 | 1 |
## Detailed Recommendations
### Features to ADD
#### 1. {Feature Name} ✅ ACCEPTED
- **Rationale**: {why}
- **Evidence**: {from research}
- **Impact**: High | **Effort**: Medium
#### 2. {Feature Name} ❌ REJECTED
- **Rationale**: {why}
- **Evidence**: {from research}
- **User Reasoning**: "{user's reason for rejection}"
### Features to REMOVE
...
### Features to MODIFY
...
## Final Priority Assessment
### P0 - Must Have
1. {feature}
2. {feature}
### P1 - Should Have
1. {feature}
### P2 - Nice to Have
1. {feature}
## Differentiation Strategy
{summary of how to position against competitors}
---
*Evaluation conducted by speckit-architect (Opus 4.5)*
*User decisions recorded: {date}*
Handoff
After evaluation is complete and user approves:
- •Update spec status to
evaluated - •Next step: Phase 4 (Design Session)
- •Spec is now frozen for design phase
Key Principles
- •User has final say - AI suggests, user decides
- •Evidence-based - Every recommendation cites research
- •No scope creep - Removing features is as valuable as adding
- •Priority clarity - Clear P0/P1/P2 before design