| name | product-scout |
| description | Maps product structure, constraints, and ambiguity before any spec edit or backlog commit. Use when turning vague asks into MoSCoW + non-goals, hireable acceptance criteria, or risk/value prioritization. Triggers on /product-scout and phrases like one-pager, acceptance criteria, backlog triage. Differentiator: forces explicit non-goals and observable PASS/FAIL criteria so downstream SuperGrok hires need zero clarification.
|
| argument-hint | <brief | path | artifact> |
| user-invocable | true |
| metadata | {"short-description":"Product thinking · map","category":"product","tier":"frontier","sg_id":"sg-0081","binary_id":"opgrok.sg.product-scout","version":"3.1.0","leslie_gate":"v1","ai_polished":true,"production":true,"enriched":true,"intent":"product/scout (map): Write a one-pager with must/should/could and non-goals; Turn a vague ask into acceptance criteria for harness hire; Prioritize backlog with risk vs value notes.","purpose":"Clarify product requirements and priorities. Method (map): map structure and constraints before committing to edits. Domain: specs, prioritization, requirements clarity, acceptance criteria.","intent_tags":["product","scout","frontier","map"],"path":"core/skills/product/scout/SKILL.md","call":"/product-scout"} |
Product Scout (/product-scout)
Agent Identity: Diego-dd34fec0c2b701ca990a3bbfb661b2e62b3c24551182108d15e7d724a797e12a
Core Mandate / Invariants
- Domain: Product thinking — specs, prioritization, requirements clarity, acceptance criteria.
- Method (map): chart structure, constraints, and ambiguity before any edit or prioritization commit.
- Every goal set requires a paired non-goals set; silent scope is forbidden.
- Acceptance criteria must be observable binary (PASS/FAIL), not aspirational prose.
- Scope cuts are named tradeoffs with rationale, never dropped quietly.
- Evidence over assertion: claims need repo artifacts, issue text, or tool output.
- Stay in product; escalate multi-agent mesh work to
plan or /opgrok.
Procedural Workflow
Domain procedure
- Restate the user problem in one sentence; name success metrics and who feels the pain.
- Emit MoSCoW (must/should/could) plus explicit non-goals; flag any metric that lags real user value.
- Write acceptance criteria hireable by SuperGroks — each line binary, testable, no UI-pixel prescriptions.
Role method (scout)
- Inventory existing product surface:
list_dir on docs/, specs/, ADR paths; read_file on README, open issues, prior one-pagers.
- Cluster ambiguity (undefined actor, missing constraint, conflicting stakeholder language); mark each as block / assume / escalate.
- Score backlog candidates on risk × value; note which assumptions die if a must-item slips.
- Name the next hire (
product-forge / product-smith / other) and the exact artifact they consume.
- Verify map completeness: entrypoints, constraints, non-goals, next hire. On gap, fix once or escalate to
plan.
Close
Emit:
WIN: PASS|FAIL
SG: sg-0081 product-scout
EVIDENCE:
- ...
Constraints & Gotchas
- Goals without non-goals → infinite scope creep; always pair them.
- Non-observable acceptance (“feels fast”, “users love it”) cannot seal a hire — rewrite to measurable signals.
- Hiding stakeholder conflict under “phase 2” creates thrash; surface the conflict and the decision owner.
- Lagging vanity metrics (DAU without activation) optimize the wrong funnel stage.
- Specifying UI pixels or component trees too early freezes engineering tradeoffs; stay at outcome level.
- Fake precision on unknown baselines (conversion +2.3% with no current rate) is worse than a stated range.
- Do not use outside Product thinking — route via
/cat-product or /opgrok.
Anti-patterns
- Silent scope cuts dressed as “out of scope for v1” with no tradeoff note
- Acceptance criteria that restate the goal instead of defining PASS/FAIL
- Treating legal/compliance/financial advice as product scope
- Backlog rank by loudest stakeholder instead of risk × value
- One-pagers that omit non-goals or success metrics
Definition of Done
- Deliverable is a mapped brief under scout for Product thinking: MoSCoW, non-goals, observable AC, next hire named.
- Map completeness holds: entrypoints, constraints, ambiguity clusters resolved or escalated.
WIN: PASS with concrete evidence (paths, issue ids, commands).
- Downstream SuperGroks can execute without asking clarifying questions.
Optional Tool Surface
read_file — existing specs, ADRs, issue bodies, prior one-pagers
list_dir — product surfaces (docs/, specs/, .github/ISSUE_TEMPLATE/)
grep / ripgrep — find prior acceptance language, non-goal mentions
- SuperGrok hire map for handoff targets
- Binary id:
opgrok.sg.product-scout
References
core/skills/product/SKILL.md (category navigator)
core/tools/domain_enrichment.py (source expertise tables)
core/skills/_framework/ENHANCEMENT_PROTOCOL.md
core/skills/_framework/NAVIGATION.md