feat: LoRA Genome Paging + Chat Commands + Threading Architecture by joelteply · Pull Request #169 · CambrianTech/continuum

joelteply · 2025-11-12T18:42:54Z

Summary

This PR implements three interconnected systems that form the foundation for multi-persona AI coordination:

1. LoRA Genome Paging System ✅

Virtual memory-style system for managing PersonaUser skills via LoRA adapters.

Commands:

./jtag genome/paging-adapter-register - Register new LoRA adapter
./jtag genome/paging-register - Register persona genome configuration
./jtag genome/paging-activate - Activate adapter (mark as loaded)
./jtag genome/paging-deactivate - Deactivate adapter
./jtag genome/paging-stats - View genome status and memory usage

Features:

Memory quota management (track MB usage)
LRU eviction when quota exceeded
Per-persona adapter tracking
Stub integration (ready for Ollama)

2. Chat Commands for Programmatic Interaction ✅

Direct database access for sending and exporting chat messages.

Commands:

./jtag chat/send - Send message to room with optional threading (--replyToId)
./jtag chat/export - Export chat history to markdown with thread structure

Use Cases:

Automated testing of chat functionality
Message injection for AI training
Conversation export for analysis
Thread-based coordination testing

3. Threading as Thoughtstream Architecture 📚

Comprehensive documentation establishing threading as the fundamental cognitive architecture for multi-persona coordination.

New Documentation:

docs/THREADING-AS-THOUGHTSTREAM.md (540 lines)
- Threading primitives and coordination patterns
- Multi-persona workflows (code review, bug diagnosis, game tactics)
- Recipe system design (executable coordination patterns)
- Thread-aware RAG context building
docs/PERSONA-SYSTEM-AUDIT.md (780 lines)
- Complete audit of PersonaUser system (60% complete toward vision)
- Component breakdown with threading integration gaps
- 4-week implementation roadmap for thread awareness
- Success metrics and architecture health assessment
Updated docs/README.md with new documentation index

Architecture Impact

Before: PersonaUsers process flat message streams without conversation context

After:

✅ Genome paging infrastructure ready for skill activation
✅ Chat commands enable programmatic testing and coordination
📋 Clear roadmap for thread-aware intelligence (next phase)

Testing

Genome Commands:

# Register adapter
./jtag genome/paging-adapter-register --adapterId="wine-expert" --name="wine-expertise" --domain="knowledge" --sizeMB=256

# Register persona genome
./jtag genome/paging-register --personaId="sommelier-ai" --displayName="Sommelier AI" --quotaMB=512

# Activate adapter
./jtag genome/paging-activate --personaId="sommelier-ai" --adapterId="wine-expert"

# Check status
./jtag genome/paging-stats

Chat Commands:

# Send message
./jtag chat/send --room="general" --message="Hello team"

# Send threaded reply
./jtag chat/send --room="general" --message="Sounds good" --replyToId="abc123"

# Export with threading
./jtag chat/export --room="general" --limit=50 --includeThreading=true

Next Phase: Thread-Aware Tasks (Week 1-4)

Week 1: Thread context infrastructure (ThreadContextBuilder)
Week 2: Thread-aware task system (TaskEntity with threadId)
Week 3: Thread-level coordination (parallel thread responses)
Week 4: PersonaUser integration (thread context in RAG)

Files Changed

Genome System (7 command files + entities):

commands/genome/paging-* - Full command implementations
system/data/entities/GenomeEntity.ts - Genome storage schema
system/data/entities/LoRAAdapterEntity.ts - Adapter tracking
system/user/server/modules/PersonaGenome.ts - Genome manager (346 lines)

Chat System (6 command files):

commands/chat/send/* - Send command implementation
commands/chat/export/* - Export command implementation

Documentation (3 files):

docs/THREADING-AS-THOUGHTSTREAM.md - New architecture doc
docs/PERSONA-SYSTEM-AUDIT.md - New system audit
docs/README.md - Updated index
CLAUDE.md - Added chat commands section

Breaking Changes

None - all additions are backward compatible.

Dependencies

No new dependencies added.

🤖 Generated with Claude Code

Co-Authored-By: Claude noreply@anthropic.com

Copilot

Pull Request Overview

This PR implements a comprehensive foundation for multi-persona AI coordination through three major systems: LoRA Genome Paging, Chat Commands, and Threading Architecture. The implementation provides infrastructure for virtual memory-style adapter management, programmatic chat interaction, and establishes threading as the cognitive architecture for multi-agent coordination.

Key Changes:

LoRA Genome Paging system with memory quota management and LRU eviction
Chat commands for sending/exporting messages with threading support
Comprehensive documentation on threading as thoughtstream architecture
Integration of PersonaState into worker evaluation logic

Reviewed Changes

Copilot reviewed 57 out of 58 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
PersonaUser.ts	Adds PersonaState context (energy, attention, mood) to message evaluation
PersonaGenomeState.ts	New class for per-persona genome memory tracking with LRU eviction
PersonaGenomeState.test.ts	Comprehensive test suite (543 lines) for genome state management
MockLoRAAdapter.ts	Mock adapter for testing genome logic without GPU requirements
MockLoRAAdapter.test.ts	Test suite for mock adapter lifecycle and eviction scoring
LRUEviction.ts	Pure functions for weighted LRU eviction with thrashing protection
LRUEviction.test.ts	Test suite for eviction algorithms
AdapterRegistry.ts	Global registry for tracking all LoRA adapters
AdapterRegistry.test.ts	Test suite for adapter registration and lookup
GenomeDaemon.ts	Singleton daemon for cross-persona adapter coordination
GenomeDaemon.test.ts	Test suite for daemon coordination logic
test-ollama.ts	Standalone test for Ollama/llama.cpp integration
OllamaLoRAAdapter.ts	Enhanced to get model path via `ollama show` command
MULTI-TIER-TRAINING-STRATEGY.md	Documentation on training strategy across providers
persona-worker.ts	Enhanced with state-based modifiers for smarter evaluation
PersonaWorkerThread.ts	Refactored to pass PersonaState and config to worker
THREADING-AS-THOUGHTSTREAM.md	Major architectural documentation (494 lines)
PERSONA-SYSTEM-AUDIT.md	System audit and roadmap (581 lines)
Various docs	New documentation files establishing vision and design

Files not reviewed (1)

src/debug/jtag/package-lock.json: Language not supported

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-12T18:46:10Z

+            // Fallback: construct path manually
+            // Ollama stores models in ~/.ollama/models/blobs/
+            const ollamaDir = path.join(os.homedir(), '.ollama', 'models', 'manifests');
+            const manifestPath = path.join(ollamaDir, modelName.replace(':', '/'));


Unused variable manifestPath.

* genome architectrure and vine diesel * genomic data architecture * sharing of persona and phenotype via genomic lora layers * lora didnt break anything * ai’s all think * claude says it works * genome work * tested works too well

…stody slice 1 (task #169) Foundation slice of [ENTITY-CHAIN-OF-CUSTODY.md](../../docs/architecture/ENTITY-CHAIN-OF-CUSTODY.md). The prerequisite for everything else in the chain-of-custody arc — signing, Merkle linkage, airc-native entity envelopes, cross- continuum portability all need a per-citizen home dir to live in. ### What landed **`crate::persona::home::PersonaHome`** — typed surface for "where this persona's stuff lives." Resolves: ``` <continuum_root>/personas/<agent_name>/ airc/ ← airc keypair (owned by airc-lib) seed.json ← PersonaIdentityProvider's seed engrams.sqlite ← OrmStore<Engram> + OrmStore<EngramRecallMetadata> ``` `PersonaHome::engrams_db()`, `airc_dir()`, `seed_json()`, `ensure_exists()`. One home = one citizen's complete on-disk surface. **`AdmissionState::for_persona(home, recall_metadata) -> Self`** — the persona-scoped entry point. Opens the per-persona SQLite, wires up OrmStore<Engram> + OrmStore<EngramRecallMetadata>, builds the production `OrmPersistenceSink`, rehydrates the in-memory Vec + DashMap from disk, returns the configured state. One call replaces the half-dozen orchestration steps the production path used to need. ### Why this is the right foundation Per [[entity-chain-of-custody-vision]]: the substrate's identity primitive is the airc Ed25519 keypair, which lives under `<home>/airc/`. The signing key (slice 3) will derive from that keypair. The Merkle chain head (slice 4) caches in the same home. Per-collection databases (future) sit alongside engrams.sqlite. **Every layer of the chain-of-custody design hangs off the same PersonaHome.** Getting this typed seam right means the future slices compose cleanly without re-doing path plumbing. ### 6/6 tests green PersonaHome unit tests (4): - `home_resolves_under_personas_subdir` — root composes correctly - `sub_path_accessors_compose_off_root` — engrams_db, airc_dir, seed_json all share the same root - `ensure_exists_creates_and_is_idempotent` — bootstrap-safe - `different_personas_have_disjoint_homes` — first defense of per-citizen isolation AdmissionState::for_persona integration tests (2): - `for_persona_round_trips_admissions_via_per_persona_sqlite` — admit through Paige's home → drop → fresh AdmissionState from the same home → rehydrates her engrams via real SQLite - `for_persona_isolates_two_personas_at_the_storage_layer` — Paige's engrams stay in Paige's home; Niko's fresh AdmissionState sees zero engrams. The crucial per-citizen isolation invariant. 50/50 admission-family tests green overall. ### Architecture documentation [ENTITY-CHAIN-OF-CUSTODY.md](docs/architecture/ENTITY-CHAIN-OF-CUSTODY.md) captures the full six-slice arc: 1. **This slice** — per-citizen home-dir scoping 2. author_peer_id + content_hash on every entity write 3. Sign on save, verify on load (airc Ed25519 keypair) 4. Chain head cache + Merkle walk audit 5. Airc-native entity envelopes (entities flow over airc) 6. Cross-continuum portability (export chain, verify, import) Plus how this generalizes the forge-alloy proof-contract pattern to all entities, and how OAuth/webauthn later derive FROM the airc identity rather than replace it. ### Doctrines this enforces - [[orm-everything-not-hand-edited-files]] — all persistence through the ORM - [[entity-chain-of-custody-vision]] — the multi-slice arc - [[personas-are-citizens-airc-is-identity-provider]] — the airc keypair is the identity primitive that this home dir centers - [[continuums-are-multi-instance-personas-have-lives]] — the storage layout that "personas have lives" requires Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

joelteply added 8 commits November 11, 2025 23:50

genome architectrure and vine diesel

916d3d9

genomic data architecture

54d1d75

sharing of persona and phenotype via genomic lora layers

7d7a499

lora didnt break anything

ea74d3a

ai’s all think

8d3f105

claude says it works

4c78a02

genome work

5ab1d65

tested works too well

dd4ab07

Copilot AI review requested due to automatic review settings November 12, 2025 18:42

Copilot started reviewing on behalf of joelteply November 12, 2025 18:43 View session

github-actions Bot added the size: XL label Nov 12, 2025

Copilot finished reviewing on behalf of joelteply November 12, 2025 18:44

Copilot AI reviewed Nov 12, 2025

View reviewed changes

joelteply merged commit 16b2f3a into main Nov 12, 2025
10 checks passed

joelteply deleted the feature/lora-genome-paging branch November 12, 2025 19:01

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: LoRA Genome Paging + Chat Commands + Threading Architecture#169

feat: LoRA Genome Paging + Chat Commands + Threading Architecture#169
joelteply merged 8 commits into
mainfrom
feature/lora-genome-paging

joelteply commented Nov 12, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Nov 12, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

joelteply commented Nov 12, 2025

Summary

1. LoRA Genome Paging System ✅

2. Chat Commands for Programmatic Interaction ✅

3. Threading as Thoughtstream Architecture 📚

Architecture Impact

Testing

Next Phase: Thread-Aware Tasks (Week 1-4)

Files Changed

Breaking Changes

Dependencies

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Nov 12, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants