Clarify catalog ownership and method boundaries

Author: raphaelmarra

skills/select-methodologies/SKILL.md

@@ -108,6 +108,8 @@ Read only the files needed:
 
 For a cross-domain problem, begin with the primary decision catalog and at most two adjacent domain catalogs. Add another only when a selected fragment is canonically owned there or a material hard gate remains uncovered. Route by decision and missing capability, not by every noun in the prompt. `references/10-composition-patterns.md` and `references/11-source-registry.md` do not count as domain catalogs. Do not load all references by default.
 
+When a prompt says only “uncertainty,” identify the required output before routing: probability/statistical inference, decision choice, optimization, causal effect, forecast, risk treatment, or evidence confidence. Route statistical/probabilistic inference to `37`; route choices, optimization, causal identification, and decision robustness to `08`. Use both only when the statistical result is an input to a distinct decision or causal design.
+
 ### 3. Build a role-aware longlist
 
 Search by problem signal, deliverable, and failure mode—not only by familiar names. Include candidates from distinct roles where warranted:

skills/select-methodologies/references/00-taxonomy-selection.md

@@ -52,6 +52,12 @@ Use this schema when adding or deeply researching a method:
 
 For a discovery catalog, the compact fields `type`, `purpose`, `use`, `avoid`, `complements`, and `source anchor` are sufficient. Expand only shortlisted methods.
 
+Write the boundaries as decision rules, not slogans:
+
+- `Use when` should name a context signal, the minimum precondition, and the output the method will produce.
+- `Do not use when` should name a failed assumption, an incompatible output, an authority/validity boundary, or a safer/appropriate alternative.
+- State a disadvantage elsewhere only when it does not determine method fit. Do not turn every cost into a prohibition.
+
 ## Context triage
 
 ### Problem topology

skills/select-methodologies/references/10-composition-patterns.md

@@ -64,6 +64,22 @@ Use one generic statistical, causal, process, or evidence card as the canonical
 
 This preserves discoverability without maintaining competing definitions. It also lets Situational Method Engineering compose a generic fragment with domain-specific gates and artifacts.
 
+### Canonical family examples
+
+| Family | Canonical owner | Secondary routing or specialization |
+| --- | --- | --- |
+| Probability, sampling, estimation, generic statistical inference | `37-probability-statistics-inference.md` | decision uncertainty in `08`; process control in `03`; spatial methods in `33`; clinical evidence in `35` |
+| Decision analysis, optimization, causal identification | `08-decision-optimization-causality.md` | finance in `12`; policy evaluation in `13`; education evaluation in `18` |
+| Theory of Change / logic model | `08-decision-optimization-causality.md` | strategy framing in `02`; policy/program evaluation in `13`; learning evaluation in `18` |
+| Design discovery and convergence | `02-strategy-product-organization.md` | design execution in `20`; innovation/R&D in `28` |
+| Experimental design / DOE | `37-probability-statistics-inference.md` | process/energy in `15`; innovation and technology maturation in `28`; agriculture in `17` |
+| API workflow description / Arazzo | `04-software-systems-architecture.md` | formal planning and orchestration in `05` |
+| Project controls / EVM | `30-project-programme-portfolio-management.md` | investment or financial appraisal in `12` |
+| Human-centred design | `20-design-experience-communication.md` | safety-critical human factors and medical devices in `35` |
+| Place-based analysis | `38-urban-territorial-planning.md` | financial appraisal in `12`; geospatial evidence in `33` |
+
+The secondary entry must state what it inherits and what changes. A repeated name without that distinction is a routing defect, even when the method is genuinely relevant in both domains.
+
 ## Reusable stack patterns
 
 ### Improve a stable process

tests/bench/2026-08-16-catalog-routing/resultados.md

@@ -0,0 +1,25 @@
+# Results
+
+## Verdict
+
+PASS — the routing and boundary review was applied without repository regression.
+
+## Evidence
+
+- `python tools/validate_repository.py`: passed; 39 catalogs and references checked.
+- `git diff --check`: passed.
+- Method-table scan: `EMPTY_USE_AVOID=0`.
+- Canonical family map: present in `references/10-composition-patterns.md`.
+- Ambiguous “uncertainty” routing: explicitly separated in `SKILL.md` between statistical inference (`37`) and decision/causal/optimization work (`08`).
+
+## Changes reviewed
+
+- Added decision-rule guidance for `Use when` and `Do not use when` in the taxonomy protocol.
+- Added canonical owners and secondary specializations for representative duplicated families.
+- Preserved the existing capability-boundary structure of the crisis catalog after confirming it already met the intended pattern.
+
+## Adversarial outcome
+
+- No empty method boundaries were introduced.
+- No reference-file or local-link validation failure was introduced.
+- The generic-versus-specialized distinction is now explicit for the tested families.