Review dimensions for validating agent quality - template compliance, safety, testing, and priority validation
The English summary is being prepared. Showing the original description.
4-phase discovery workflow with decision gates, phase transitions, success metrics, and state tracking
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
PHASE 1 PHASE 2 PHASE 3 PHASE 4
Problem Validation Opportunity Mapping Solution Testing Market Viability
| | | |
v v v v
"Is this real?" "Which matters?" "Does it work?" "Viable business?"
Duration: 1-2 weeks | Min interviews: 5 | Techniques: Mom Test, Job Mapping Core question: Is this a real problem worth solving?
Duration: 1-2 weeks | Min interviews: 10 cumulative | Techniques: OST, Opportunity Algorithm Core question: Which problems matter most?
Duration: 2-4 weeks | Min interviews: 5 per iteration | Techniques: hypothesis testing, prototypes Core question: Does our solution actually work?
Duration: 2-4 weeks | Min interviews: 5 + stakeholders | Techniques: Lean Canvas, 4 Big Risks Core question: Can we build a viable business?
Proceed: 5+ confirm pain + willingness to pay | Pivot: problem differs from expected | Kill: <20% confirm
Proceed: top 2-3 score >8 (max 20) | Pivot: new opportunities discovered | Kill: all low-value Scoring: Score = Importance + Max(0, Importance - Satisfaction). Each 1-10. >8 = high importance with satisfaction gap.
Proceed: >80% task completion, usability validated | Pivot: needs refinement | Kill: fundamental blocks
Proceed: all 4 risks addressed, model validated | Pivot: model needs adjustment | Kill: no viable model
| Metric | Target |
|---|---|
| Problem confirmation | >60% (3+ of 5) |
| Frequency | Weekly+ |
| Current spending | >$0 on workarounds |
| Emotional intensity | Frustration evident |
Done when: 5+ interviews, >60% confirmation, articulated in customer words, 3+ examples. Threshold: 60% aligns with Mom Test -- 3/5 consistent signals = proceed, <20% = kill. Combined with qualitative markers.
| Metric | Target |
|---|---|
| Opportunities identified | 5+ distinct |
| Top scores | >8 / max 20 |
| Job step coverage | 80%+ |
| Strategic alignment | Stakeholder confirmed |
Done when: OST complete, top 2-3 prioritized, team aligned.
| Metric | Target |
|---|---|
| Task completion | >80% |
| Value perception | >70% "would use/buy" |
| Comprehension | <10 sec to understand |
| Key assumptions validated | >80% proven |
Done when: 5+ users per iteration, core flow usable, value + feasibility confirmed.
| Metric | Target |
|---|---|
| Four big risks | All green/yellow |
| Channel validated | 1+ viable |
| Unit economics | LTV > 3x CAC (estimated) |
| Stakeholder signoff | Legal, finance, ops |
Done when: Lean Canvas complete, all risks acceptable, go/no-go documented.
current_phase: "1|2|3|4"
phase_started: "ISO timestamp"
interviews_completed: "count by phase"
assumptions_tracked: "list with risk scores"
opportunities_identified: "list with scores"
decision_gates_evaluated: "G1|G2|G3|G4 status"
artifacts_created: "list of file paths"
| Transition | Gate | Requirements |
|---|---|---|
| 1 to 2 | G1 | 5+ interviews, >60% confirmation, customer words, 3+ examples |
| 2 to 3 | G2 | OST complete, top 2-3 identified, scores >8, team alignment |
| 3 to 4 | G3 | 5+ users tested, >80% task completion, core flow usable, validated |
| 4 to handoff | G4 | Lean Canvas complete, 4 risks acceptable, go/no-go, stakeholder sign-off |
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Review dimensions for validating agent quality - template compliance, safety, testing, and priority validation
The English summary is being prepared. Showing the original description.
Review dimensions for validating agent quality - template compliance, safety, testing, and priority validation
The English summary is being prepared. Showing the original description.
Review dimensions for acceptance test quality - happy path bias, GWT compliance, business language purity, coverage completeness, walking skeleton user-centricity, priority validation, observable behavior assertions, traceability coverage, and walking skeleton boundary proof
The English summary is being prepared. Showing the original description.
Detailed 5-phase workflow for creating agents - from requirements analysis through validation and iterative refinement
The English summary is being prepared. Showing the original description.
5-layer testing approach for agent validation including adversarial testing, security validation, and prompt injection resistance
The English summary is being prepared. Showing the original description.
Architectural style selection decision matrices, trade-off analysis, structural enforcement rules, and combination patterns. Load when choosing or evaluating architecture styles.
The English summary is being prepared. Showing the original description.