Lyra Core — Project Summary

v0.4 (2025-10-03)

Lyra Core (v0.3.1) — Orchestration layer.
- Accepts chat requests (/v1/chat/completions).
- Routes through Cortex for subconscious annotation.
- Stores everything in Mem0 (no discard).
- Fetches persona + relevant memories.
- Injects context back into LLM.
Cortex (v0.3.0) — Subconscious annotator.
- Runs locally via llama.cpp (Phi-3.5-mini Q4_K_M).
- Strict JSON schema:
```
{
  "sentiment": "positive" | "neutral" | "negative",
  "novelty": 0.0–1.0,
  "tags": ["keyword", "keyword"],
  "notes": "short string"
}
```
- Normalizes keys (lowercase).
- Strips Markdown fences before parsing.
- Configurable via .env (CORTEX_ENABLED=true|false).
- Currently generates annotations, but not yet persisted into Mem0 payloads (stored as empty {cortex:{}}).
Mem0 (v0.4.0) — Persistent memory layer.
- Handles embeddings, graph storage, and retrieval.
- Dual embedder support:
  - OpenAI Cloud (text-embedding-3-small, 1536-dim).
  - HuggingFace TEI (gte-Qwen2-1.5B-instruct, 1536-dim, hosted on 3090).
- Environment toggle for provider (.env.openai vs .env.3090).
- Memory persistence in Postgres (payload JSON).
- CSV export pipeline confirmed (id, user_id, data, created_at).
Persona Sidecar
- Provides personality, style, and protocol instructions.
- Injected at runtime into Core prompt building.

Mem0
- Added HuggingFace TEI integration (local 3090 embedder).
- Enabled dual-mode environment switch (OpenAI cloud ↔ local TEI).
- Fixed .env line ending mismatch (CRLF vs LF).
- Added memory dump/export commands for Postgres.
Core/Relay
- No major changes since v0.3.1 (still routing input → Cortex → Mem0).
Cortex
- Still outputs annotations, but not yet persisted into Mem0 payloads.

Wire Cortex annotations into Mem0 payloads (cortex object).
Add “export all memories” script to standard workflow.
Consider async embedding for faster mem.add.
Build visual diagram of data flow (Core ↔ Cortex ↔ Mem0 ↔ Persona).
Explore larger LLMs for Cortex (Qwen2-7B, etc.) for richer subconscious annotation.