skills: state reference-file location in every References section

Each SKILL.md now says reference files live in references/ at the plugin
root (${CLAUDE_PLUGIN_ROOT}/references/ when installed), making retrieval
deterministic instead of best-effort filename hunting.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: geledek

skills/general-idea-diagnostic/SKILL.md

@@ -127,6 +127,8 @@ See `cases/` for diagnosed examples. When diagnosing a new idea, compare against
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `95-5-genai-divide.md` — empirical basis for why rigorous gating matters: 95% of pilots produce zero measurable P&L impact
 - `european-fintech-case.md` — canonical measurement-gap failure case
 - `mckinsey-3-objective-mix.md` — objective-mode classification (efficiency / growth / innovation)

skills/general-maturity-assessment/SKILL.md

@@ -145,6 +145,8 @@ CURRENT STATE SUMMARY | TARGET STATE | CAPABILITY GAPS (ordered) | BOLD GOAL | R
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `mit-cisr-4-stages.md` — four-stage model; performance data; Stage 3 inflection; DBS/Ping An exemplars
 - `accenture-maturity-archetypes.md` — Foundation × Differentiation 2×2; archetype definitions; Achiever performance premium
 - `mckinsey-3-objective-mix.md` — efficiency/growth/innovation objective mix

skills/general-peer-cases/SKILL.md

@@ -145,6 +145,8 @@ PROFILE | CASE BUNDLE | CROSS-CASE LESSONS | CLOSEST-MATCH ACTION | CONFIDENCE |
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `stanford-51-deployments.md` — 51-deployment study; 77% invisible-cost finding; canonical retrieved-case set.
 - `95-5-genai-divide.md` — base-rate anchor: 95% of pilots produce zero measurable P&L impact; 5% breakthrough criteria.
 - `european-fintech-case.md` — canonical Replace-mode case: 700 agents replaced, CSAT -22%, hiring resumed; the measurement-gap failure mode.

skills/general-roi-gate/SKILL.md

@@ -119,6 +119,8 @@ RECOMMENDATION | CONDITIONS (if any) | RED FLAGS (if any) | VERDICT RATIONALE
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `pwc-20-item-checklist.md` — the full 20-item Do's and Don'ts list
 - `pwc-roi-2026-governance.md` — 1.6× vs. 2.6× productivity gap; 1.7× RAI governance advantage
 - `mckinsey-3-objective-mix.md` — efficiency / growth / innovation objective classification

skills/general-use-case-discovery/SKILL.md

@@ -101,6 +101,8 @@ Default stance: if fewer than 3 candidates clear the bar, return what you have a
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - mckinsey-3-objective-mix.md — productivity / growth / transformation portfolio balance, ~50/30/20 default.
 - andrew-ng-three-moats.md — proprietary data, workflow embedding, distribution as the only durable moats in applied AI.
 - pilot-discipline-ng.md — narrow + reversible + measurable + owned filter for first-pilot selection.

skills/people-frontline-engagement/SKILL.md

@@ -169,6 +169,8 @@ VERDICT | ENGAGEMENT PLAN | FOUR-COLUMN TASK SPLIT | SAFETY CONTRACT | TRADECRAF
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `stanford-51-deployments.md` — 77% of deployment costs are invisible (change, trust, redesign); tradecraft preservation as first-class deliverable
 - `imda-4-dimensions-agentic.md` — named human accountability (Dimension 2); structural over prompt-layer controls
 - `bcg-manager-modeling.md` — leader-speaks-last norm; manager modeling as the unlock for frontline adoption

skills/people-literacy-curriculum/SKILL.md

@@ -155,6 +155,8 @@ Verdict states:
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - eu-ai-act-essentials.md — Art. 4 mandatory literacy duty, scope, refresh, evidence requirements.
 - bcg-manager-modeling.md — 88/25 finding; manager weekly use as the adoption multiplier.
 - deloitte-cheerleader-to-champion.md — 30% champion-level executive sponsorship as ceiling on adoption.

skills/people-readiness-conversation/SKILL.md

@@ -134,6 +134,8 @@ Consolidate Roles 1–4. Write for CEO or CHRO audience.
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `70-20-10-value-split.md` — BCG 70% people value; context for why these gaps matter
 - `bcg-manager-modeling.md` — 88/25 role-modeling gap; specific manager behaviors
 - `mckinsey-3x-senior-ownership.md` — 3× CEO ownership differential

skills/process-pilot-design/SKILL.md

@@ -154,6 +154,8 @@ Produce a one-page pilot brief from Roles 1–6. Write for sponsor approval. Be
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `95-5-genai-divide.md` — Scope × Execution Grid; 90-day timeline; narrow-scope success patterns
 - `stanford-51-deployments.md` — invisible-costs; workflow-redesign-first; pre-deployment metrics discipline
 - `european-fintech-case.md` — measurement-gap reference case

skills/process-portfolio-observability/SKILL.md

@@ -82,6 +82,8 @@ KPI_TREE | DASHBOARD_SPEC | REVIEW_CADENCE | SUNSET_CRITERIA | STATE: Observable
 
 ## References
 
+*All files below live in `references/` at the plugin root (`${CLAUDE_PLUGIN_ROOT}/references/` when installed as a plugin).*
+
 - `pwc-roi-2026-governance.md` — net-of-oversight ROI as the disclosure-grade number; gross ROI no longer defensible.
 - `pwc-20-item-checklist.md` — 20-item governance audit; items 11–16 cover input/output/outcome KPI wiring.
 - `imda-4-dimensions-agentic.md` — Dimension 4 structural controls (bounded tools, bounded budget, kill-switch) maps to dashboard guardrails.