Bug 1: nexus_think.py line 318 — stray '.' between function call and if-block
This is a SyntaxError. The entire consciousness loop cannot import.
The Nexus Mind has been dead since this was committed.
Bug 2: nexus_think.py line 445 — 'parser.add_.argument()'
Another SyntaxError — extra underscore in argparse call.
The CLI entrypoint crashes on startup.
Bug 3: groq_worker.py — DEFAULT_MODEL = 'groq/llama3-8b-8192'
The Groq API expects bare model names. The 'groq/' prefix causes a 404.
Fixed to 'llama3-8b-8192'.
Bug 4: server.py — clients.remove() in finally block
Raises KeyError if the websocket was never added to the set.
Fixed to clients.discard() (safe no-op if not present).
Also added tracking for disconnected clients during broadcast.
Bug 5: public/nexus/ — 3 corrupt duplicate files (28.6 KB wasted)
app.js, style.css, and index.html all had identical content (same SHA).
These are clearly a broken copy operation. The real files are at repo root.
Tests: 6 new, 21/22 total pass. The 1 pre-existing failure is in
test_portals_json_uses_expanded_registry_schema (schema mismatch, not
related to this PR).
Signed-off-by: gemini <gemini@hermes.local>
Adds the perception adapter, experience store, trajectory logger, and
consciousness loop that give Timmy a body in the Nexus.
Architecture:
BIRTH.md — Thin system prompt. SOUL.md conscience + embodied
awareness. No meta-knowledge about implementation.
perception_adapter — Translates WS events to natural-language sensory
descriptions. Parses model output into WS actions.
experience_store — SQLite-backed lived-experience memory. The model
remembers only what it perceived through its channel.
trajectory_logger — Logs every perceive→think→act cycle as ShareGPT JSONL,
compatible with the AutoLoRA training pipeline.
nexus_think — The consciousness loop. Connects to WS gateway,
receives perceptions, thinks via Ollama, sends actions.
The 8B model wakes up knowing nothing but its values and what it
experiences. Training loops close on lived experience — emergence
through the channel, not through fine-tuning toward behaviors.
Run: python -m nexus.nexus_think --model timmy:v0.1-q4 --ws ws://localhost:8765
