fix: Truth reveal flow - Hermes decodes when asked

- Updated detective-play skill
- Hermes decodes truth.enc when user asks (not user)
- Added statistics display after truth reveal
- Investigation phase stays spoiler-free
- Updated detective-create skill

docs/skills/detective-create/SKILL.md

@@ -14,9 +14,9 @@ Create a folder in `~/.hermes/detective/cases/<case-name>/`
 Each case needs:
 ```
 <case-name>/
-├── case.yaml      # Case data
+├── case.yaml      # Case data (evidence, suspects)
 ├── truth.enc      # Encoded truth
-└── images/        # Evidence images
+└── images/       # Evidence images
 ```
 
 ## Case YAML Format
@@ -34,11 +34,6 @@ evidence:
     image: "images/evidence-01.jpg"
     description: "Brief description"
 
-  - id: "evidence-02"
-    name: "Another Evidence"
-    image: "images/evidence-02.jpg"
-    description: "Brief description"
-
 suspects:
   - id: "suspect-01"
     name: "Suspect Name"
@@ -63,8 +58,12 @@ key_points:
 
 encoded = base64.b64encode(truth.encode()).decode()
 # Write encoded to truth.enc
+with open('truth.enc', 'w') as f:
+    f.write(encoded)
 ```
 
+The truth file is loaded AFTER investigation, not during. This keeps the mystery intact.
+
 ## Creating Images
 
 Images can be:
@@ -80,3 +79,4 @@ Images can be:
 - Case should be solvable with the evidence provided
 - Truth should align with evidence (no hidden info)
 - Include 2-4 evidence items and 2-3 suspects for Easy
+- key_points in truth should match what users can discover from evidence