rhzVSrSSSS.r"SS5r\"5rS\S\S \\\44S jrg ) z\ DEEP SELF-MODEL AWARENESS Provides accurate introspection about Eden's actual architecture a ACTUAL MEMORY ARCHITECTURE: 1. **Episodic Memory** (Context Manager) - Implementation: Python deque in eden_context_manager.py - Storage: In-memory, last 20 messages per user - Retrieval: Sequential access, LRU eviction - Verification: Can re-read conversation history 2. **Semantic Memory** (Fact Database) - Implementation: SQLite database (eden_facts.db) - Storage: Extracted facts from conversations - Retrieval: SQL queries with LIKE matching - Schema: id, category, subject, fact, timestamp, user_id 3. **Long-term Memory** (Memory Manager) - Implementation: Vector database (FAISS/ChromaDB) - Storage: 1,431+ embedded memories - Retrieval: Semantic similarity search - Types: Experiences, learnings, interactions a REASONING ARCHITECTURE: 1. **Pattern Reasoning** (eden_reasoning.py) - Handles: Math sequences, logical fallacies - Pre-processes questions before LLM - Returns structured solutions 2. **Rubicon Reasoning** (eden_rubicon_reasoning.py) - Handles: Paradoxes, ethical conflicts, physics - Advanced constraint enforcement - Meta-logical analysis 3. **Base LLM Reasoning** (Ollama/Llama) - Model: Variable (llama3.2, mistral, etc.) - Processes general queries - Enhanced by above modules u HOW I VERIFY KNOWLEDGE: 1. **Explicit Storage** (High Confidence) - If in eden_facts.db → "I learned this from you" - If in memory_manager → "I remember this experience" 2. **Episodic Recall** (Medium Confidence) - If in conversation history → "You mentioned this earlier" - Can verify by re-reading context 3. **Generative** (Low Confidence) - If not stored anywhere → "Based on my training..." - Cannot verify, only infer 4. **Limitations**: - No meta-cognitive calibration layer yet - Cannot quantify uncertainty numerically - No confidence scores on retrievals - Known gap: distinguishing "certain" vs "probable" )memory_systemsreasoning_systemsverification_methodsc<\rSrSrSrS\S\4SjrS\S\4SjrSrg ) SelfModelAwarenessHz#Provides accurate self-description.messagereturnc^UR5m[U4SjS55(ag[U4SjS55(ag[U4SjS55(ag g ) z'Detect what type of self-query this is.c3,># UH oT;v M g7fN.0q msg_lowers 6/home/james-whalen/eden-agi-project/eden_self_model.py 7SelfModelAwareness.detect_self_query..OsK'J!I~'J)zhow do you knowz how you know epistemologyc3,># UH oT;v M g7fr r rs rrrQs['Z!I~'Zr)z your memoryzyour architecturez your actual architecturec3,># UH oT;v M g7fr r rs rrrSsQ'P!I~'Pr) verificationverify calibrationrN)lowerany)selfrrs @rdetect_self_query$SelfModelAwareness.detect_self_queryKsPMMO K'JK K K! ['Z[ [ [! Q'PQ Q Q! query_typecv[S[SS-[S-[SS.nURUS5$)z)Get relevant architectural documentation.rr r)rrr)ARCHITECTURE_DOCUMENTATIONget)rr#docss rget_documentation$SelfModelAwareness.get_documentationXsU77MN67GH4ORlnASBB67MN  xx B''r"r N) __name__ __module__ __qualname____firstlineno____doc__strr r*__static_attributes__r r"rrrHs+-   (C(C(r"rr base_promptr c[RU5nU(a$[RU5nUSUSUS3nUS4$US4$)z#Add self-model awareness to prompt.z z IMPORTANT: Use this ACTUAL architectural information to answer questions about yourself. Be specific about your real implementation, not hypothetical systems. User question: r%TF) self_modelr r*)rr3r#r)enhanceds renhance_with_self_modelr7dsg--g6J++J7#}% y ~  r"N)r0r'rr5r1tupleboolr7r r"rr:sZ ,&U?B((4 ! SsuS$Y?Or"