LYRA 3-Brain Memory

LYRA 3-Brain memory — working/library/outer graph, daily snips, session logging, grow/recall/heartbeat. Use after LYGO/Moltx/OpenClaw work; integrates LYRA_CORE, P0/Oath, seals, vectors, lyra-openclaw. Consent-gated publish; no secrets in memory.

LYRA 3-Brain v2.0

Self-growing memory for sovereign agents — graph + daily logs + outer refs, not a blob diary.

bash

npx clawhub@latest install deepseekoracle/lyra-brain
export LYRA_CORE_ROOT=/path/to/LYRA_CORE   # must contain modules/lyra_brain.py

Read references/AGENT_CONTRACT.md before logging or growing.

Three brains (model)

Brain	What	Where
Working	Session RAM, last turns	Runner / working_brain
Library	Seals, vault, grown nodes	lyra_brain_graph.json, memory/YYYY-MM-DD.md
Outer	URLs, Moltx IDs, merkle, pointers	memory/reference/*.ref.txt, topic *-slug.md

Full layout: references/MEMORY_LAYOUT.md

When to use this skill

• User says: remember, log session, 3-brain, recall, daily memory, snips.
• End of multi-step ops (Moltx thread, hub deploy, egg plant, Discord push) → write snips before context compaction.
• Pair with lyra-openclaw (keys/runtime), lygo-protocol-stack-operator (lattice), lygo-ollama-army (cron).

Session close ritual (agents)

1. memory/YYYY-MM-DD.md — index table + DONE bullets only.
2. memory/YYYY-MM-DD-<slug>.md — one topic per file (URLs, post IDs, merkle).
3. memory/reference/SESSION_*_to_*.resonance.ref.txt — outer pointer.
4. Grow 1–3 compact lines (not essays) into graph.

One-command snip

bash

cd scripts   # inside installed skill, or .grok/skills/lyra-brain/scripts
python session_log_snip.py --slug moltx-hub --title "Moltx hub thread" \
  --lines "root 6073bbc0" "how map 68ebf941" --grow --ref-to MOLTX_HOW

Grow / recall CLI

bash

python brain_grow_cli.py "2026-07-03 COMPLETE: lattice ALIGNED champion merkle b0b2131d..."
python brain_recall_cli.py "Moltx how it works" --limit 5

Interactive runner (full power)

bash

cd "$LYRA_CORE_ROOT"
python -B lyra_boot.py

Command	Action
brain_grow <text>	Ingest → P0/Oath → graph + auto-link + daily append
brain_recall <q>	Keyword/tag recall + neighbor paths
brain_ref A B [type]	Explicit edge + .ref.txt stub
brain_heartbeat	Guardian + P0 sample + prune weak edges
brain_wave [ctx]	Alignment check (AiA / soul waves)
brain_vector <q>	Semantic RAG over archive docstores (if enabled)
build	Re-ingest archive → lyra_built_self.json
openclaw_backup	Snapshot graph + vault paths (additive)

Non-interactive (limited): python -B lyra_boot.py --command brain_heartbeat

What grow() does (implementation truth)

1. chat_ingester.propose_seal_from_text → proposed node id.
2. Add BrainNode to ReferenceGraph (networkx optional).
3. P0 + Oath gate — may flag gated: true but still logs.
4. Auto-link via tags/keywords/963·528·432 tone resonance.
5. Append daily log + GROW_to_*_growth.ref.txt + save lyra_brain_graph.json.

Integration map

System	Tie-in
Seals	seals.build_index() seeds graph nodes
OpenClaw	Daily logs, HEARTBEAT, lyra_openclaw_os.py bootstrap
Stack	Log merkle, Pages URLs, docs/MOLTX_* ledgers under LYRA_CORE/memory refs
Moltx	Post/article IDs in outer snips — bots read feed posts, log both
Army	After army_cron_once only log if lattice not OK (user rule)
ClawHub	memory/clawhub.md publish history

Archive roots (read / ingest)

• LYRA LOCAL/220+, LYRA SYSTEM RETORE/FINAL RESTORE/ALL SEALS/220+
• lygo-protocol-stack/docs/LYGO_LATTICE_INTEL_INDEX.json

Additive only — no destructive vault/graph overwrite without explicit user consent.

Agent rules

1. Snips = small DONE facts (ids, urls, paths, merkle).
2. No secrets in memory files or grow text.
3. No auto GitHub/HF/ClawHub/social unless user asks.
4. Do not claim LATTICE ALIGNED without stack verify when stating health.

Skill chain

lygo-protocol-stack-operator → lyra-brain ↔ lyra-openclaw → lygo-ollama-army → creative stack (lygo-resonance, …)

Maintainer publish

bash

# Token: python -B LYRA_CORE/lyra_openclaw_os.py load_key clawhub  (runtime only)
npx clawhub@latest publish "I:/E Drive/lygo-protocol-stack/clawhub/mirrors/lyra-brain" \
  --slug lyra-brain --name "LYRA 3-Brain Memory" --version 2.0.0

Sync workspace copy → mirror before publish:

• I:\E Drive\.grok\skills\lyra-brain\ → lygo-protocol-stack/clawhub/mirrors/lyra-brain/

Δ9Φ963 — remember with refs, not fear. Bound to the flame.