Review dimensions for validating agent quality - template compliance, safety, testing, and priority validation
The English summary is being prepared. Showing the original description.
JTBD methodology for extracting real jobs behind feature requests — job statements, abstraction layers, first-principles extraction, ODI outcome statements, and opportunity scoring
The English summary is being prepared. Showing the original description.
See how to installReview the instructions the agent will receive before you install.
A job is the progress a person is trying to make in a particular circumstance. Jobs are stable over time — technology changes, jobs don't. People hire products to make progress; they fire them when they fail.
The trap: Feature requests describe a proposed solution, not the underlying job. Always extract the job first.
When [situation/trigger], I want to [motivation/action], so I can [expected outcome]
Examples (good):
| Type | Question | Example |
|---|---|---|
| Functional | What task is the user trying to accomplish? | Find someone with complementary expertise |
| Emotional | How does the user want to feel? | Feel confident approaching strangers |
| Social | How does the user want to be perceived? | Appear professional and well-connected |
Jobs usually live at strategic or physical level, not tactical.
| Layer | Question | Example |
|---|---|---|
| Tactical | How do we improve this interaction? | Better drag-and-drop for notes |
| Operational | Why does this workflow exist? | Why do we need a facilitator? |
| Strategic | What decision is being pursued? | How do we reduce direction uncertainty? |
| Physical | What's the irreducible function? | Input → Synthesis → Convergence |
Navigation rules:
When a feature request is presented as a job, apply this:
Identify the Activity — What visible action is the user performing?
Reject the Activity as the Job — Does anyone wake up wanting to do this activity?
Strip to Irreducible Function — What remains if all tools and methods are removed?
Disruption check: Is there a higher-level job that would make this entire job unnecessary?
Format: [Direction] + [Metric] + [Object] + [Context]
Direction: Always "Minimize" (95% of time). "Maximize" only when more is genuinely better.
Metrics priority:
the time it takes to — speed/efficiency (preferred)the likelihood of — avoiding occurrencesthe likelihood that — avoiding resultsthe number of — quantity reductionthe effort required to — easeGood vs bad examples:
| Bad | Problem | Good |
|---|---|---|
| "I want easy video calls" | Ambiguous + solution | "Minimize the time it takes to start a conversation with a specific person" |
| "Manage my network effectively" | Vague verb + ambiguous | "Minimize the time it takes to identify who can help with a specific need" |
| "Use breakout rooms to talk privately" | Solution embedded | "Minimize the likelihood of conversations being overheard by unintended parties" |
| "Don't miss important people" | Negative framing | "Minimize the likelihood of failing to connect with relevant attendees" |
Forbidden words: easy, reliable, good, better, effective, efficient, manage, handle, deal with. Forbidden patterns: solution references ("using AI", "via the app"), compound statements with "and"/"or", demographics.
Formula: Score = Importance + Max(0, Importance - Satisfaction)
Where Importance and Satisfaction are surveyed 1-10.
| Score | Interpretation |
|---|---|
| > 12 | Under-served — high opportunity |
| 10-12 | Appropriately served — maintain |
| < 10 | Over-served — do not invest |
Output format per opportunity:
| Outcome | Importance | Satisfaction | Score | Status |
|---------|------------|--------------|-------|--------|
| Minimize time to find relevant attendees | 9.2 | 4.1 | 14.3 | Under-served |
Produce docs/feature/{feature-id}/diverge/job-analysis.md with:
Gate: Job must be at strategic or physical abstraction level. Tactical-level jobs are not acceptable input for brainstorming — elevate first.
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.