feat: add server rule violation detection and compress prompts

- LLM now evaluates messages against numbered server rules and reports
  violated_rules in analysis output
- Warnings and mutes cite the specific rule(s) broken
- Rules extracted to prompts/rules.txt for prompt injection
- Personality prompts moved to prompts/personalities/ and compressed
  (~63% reduction across all prompt files)
- All prompt files tightened: removed redundancy, consolidated Do NOT
  sections, trimmed examples while preserving behavioral instructions

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

utils/llm_client.py

@@ -92,6 +92,11 @@ ANALYSIS_TOOL = {
                     "type": ["string", "null"],
                     "description": "The game channel name this message is about (e.g. 'gta-online', 'warzone'), or null if not game-specific.",
                 },
+                "violated_rules": {
+                    "type": "array",
+                    "items": {"type": "integer"},
+                    "description": "Rule numbers violated (empty array if none).",
+                },
             },
             "required": ["toxicity_score", "categories", "reasoning", "off_topic", "topic_category", "topic_reasoning", "coherence_score", "coherence_flag"],
         },
@@ -190,6 +195,11 @@ CONVERSATION_TOOL = {
                                 "type": ["string", "null"],
                                 "description": "The game channel name this user's messages are about, or null.",
                             },
+                            "violated_rules": {
+                                "type": "array",
+                                "items": {"type": "integer"},
+                                "description": "Rule numbers violated (empty array if none).",
+                            },
                         },
                         "required": ["username", "toxicity_score", "categories", "reasoning", "off_topic", "topic_category", "topic_reasoning", "coherence_score", "coherence_flag"],
                     },
@@ -299,12 +309,15 @@ class LLMClient:
     async def analyze_message(
         self, message: str, context: str = "", user_notes: str = "",
         channel_context: str = "", mention_context: str = "",
+        rules_context: str = "",
     ) -> dict | None:
         user_content = f"=== RECENT CHANNEL MESSAGES (for background context only) ===\n{context}\n\n"
         if user_notes:
             user_content += f"=== NOTES ABOUT THIS USER (from prior analysis) ===\n{user_notes}\n\n"
         if channel_context:
             user_content += f"=== CHANNEL INFO ===\n{channel_context}\n\n"
+        if rules_context:
+            user_content += f"=== SERVER RULES ===\n{rules_context}\n\n"
         if mention_context:
             user_content += f"=== USER REPORT (a user flagged this conversation — focus on this concern) ===\n{mention_context}\n\n"
         user_content += f"=== TARGET MESSAGE (analyze THIS message only) ===\n{message}"
@@ -382,6 +395,8 @@ class LLMClient:
 
         result.setdefault("note_update", None)
         result.setdefault("detected_game", None)
+        if not isinstance(result.get("violated_rules"), list):
+            result["violated_rules"] = []
 
         return result
 
@@ -490,6 +505,7 @@ class LLMClient:
         user_notes_map: dict[str, str] | None = None,
         new_message_start: int | None = None,
         user_aliases: str = "",
+        rules_context: str = "",
     ) -> dict | None:
         """Analyze a conversation block in one call, returning per-user findings."""
         if not messages:
@@ -506,6 +522,8 @@ class LLMClient:
                 user_content += "=== USER NOTES (from prior analysis) ===\n" + "\n".join(notes_lines) + "\n\n"
         if channel_context:
             user_content += f"=== CHANNEL INFO ===\n{channel_context}\n\n"
+        if rules_context:
+            user_content += f"=== SERVER RULES ===\n{rules_context}\n\n"
         if mention_context:
             user_content += f"=== USER REPORT (a user flagged this conversation — focus on this concern) ===\n{mention_context}\n\n"
         user_content += "Analyze the conversation block above and report findings for each user."
@@ -587,6 +605,8 @@ class LLMClient:
             finding.setdefault("coherence_flag", "normal")
             finding.setdefault("note_update", None)
             finding.setdefault("detected_game", None)
+            if not isinstance(finding.get("violated_rules"), list):
+                finding["violated_rules"] = []
         result["user_findings"] = findings
         result.setdefault("conversation_summary", "")
         return result