Skip to content

INDEX.md

Latest commit

 

History

History
124 lines (101 loc) · 6.89 KB

INDEX.md

File metadata and controls

124 lines (101 loc) · 6.89 KB
Raw

CosySim Documentation Index

v1.57.2 — Gemini Native + Unified CLI. 16 CLI commands, 15 standalone apps, 7-tier query pipeline, 92 scheduler tasks, Three-Pillar Architecture + ARGUS Recon Framework, 1,040+ Skills.

Quick Reference

Metric Value
Version v1.57.2 [2026-03-27]
Targets 35 (18 game + 11 service + 6 creation)
Skills ~1,040 across 99 packs via @skill decorator
Interceptors 36 pipeline hooks in the agent governance layer
MCP Tools 43 domain modules
ARGUS 21 toolkit functions, 13 recon techniques
Tests 417 test files (smart runner: ~53s smoke)
Pillars Game (NeonCity) / Service / Creation

Three-Pillar Architecture

All targets defined in engine/control_plane_registry.py:

Pillar Count Key Targets
Game 15 phone, penthouse, lounge, tavern, casino, gallery, arena, realm, neoncity, coders, heist, games, grid, lab_break, oracle
Service 11 nexus_kms, hub, nexus_panel, dashboard, admin, tts, bridge, nlm_proxy, system_control, command_center, intel_hub
Creation 7 canvas, canvas_api, assets, creator, asset_studio, creation_kit, neonos

Key Ports

Port Service Managed
1234 LMStudio (local inference) External
5555–5572 Game scenes Auto
5570–5580 Service scenes Auto
5590–5595 Creation tools Auto
8500 Hub (scene launcher) Auto
8501–8504 Streamlit apps (dashboard, admin, assets, creator) Auto
8600 TTS Server Launcher
8700 Nexus KMS (auto-start, priority 0) Auto
5800 Model Proxy (OpenAI + Anthropic + Gemini, 21 models) Manual
5593 Advanced Assistant (chat UI + proxy) Manual
8800 NLM Proxy Auto

Architecture and Design

Doc Description
Architecture System design, layers, data flow, three-pillar targets, singletons
MCP Framework Skill dispatch, governance, state coordination, dialog system
Interceptors 36 pre/post-call hooks, priorities, auto-registry, scene filtering
Configuration YAML config hierarchy, get_config() pattern, environment overrides
Decisions 14 architecture decision records with rationale

Scenes and Content

Doc Description
Scenes All 32 targets, mechanics, APIs, routes, Socket.IO events
Skills @skill decorator, 95 packs, runtime registry, governance filtering
Game Systems WorldSim, factions, NPCs, missions, inventory, cyberspace, and the v1.62 Living City comms/phone systems (GlobalCommsLog, NPC↔NPC messaging, phone hacking, Oracle comms)
Character System Profiles, personality, stats, neurochemistry, relationships
Economy Guide EconomyManager, cross-scene credits, markets, consequences
Arena Guide THE COLOSSEUM, card game mechanics, betting
The Grid THE GRID, underground marketplace, 4 zones
Neon HUD Universal HUD v2, glass panels, phone overlay, announcer
OpenRoom Features 6 OpenRoom-inspired features: memory, danmaku, NeonOS, virtual FS, narrative, character creator

Knowledge and Integration

Doc Description
Nexus Nexus KMS v1.5.0, agent registry, KnowledgePipeline, 7-tier query router (+ Gemini File Search), NLM integration, training flywheel
Nexus System Nexus internals, 7-tier query pipeline, Gemini integration, self-maintenance, governance, agent registry
Nexus API Reference Complete Nexus REST API + Gemini File Search/structured output/context cache APIs
LMStudio InferenceOrchestrator, ServerController, LMLink federation, SSE streaming
ARGUS Browser automation, CDP, API surface discovery, RPC registry
ARGUS Methodology 13 reusable recon techniques: heap mining, bundle decompilation, flag injection, CDP scripting, agent extraction
ARGUS Discovery Journal Narrative of reverse-engineering Sesame AI + OpenRoom.ai — timeline, techniques born, insights
ARGUS: Sesame AI Report Complete intelligence: 53 API methods, WebRTC, Statsig, RLHF pipeline, security assessment
ARGUS: OpenRoom Report Complete intelligence: 5-brand mapping, multi-agent architecture, virtual OS, protobuf protocol

Infrastructure

Doc Description
Apps & CLI Unified CLI (cli.py), 15 standalone apps, venv auto-bootstrap, command reference
Operations Launcher, TUI, ports, logging, monitoring, scheduler, admin panels
TTS Qwen3-TTS server, voice design, streaming audio
Asset Studio ComfyUI integration, workflows, tuning engine
Training Data flywheel, DataCollector, AutoTrain, LoRA fine-tuning, benchmarks

Tools and SDKs

Doc Description
Advanced Assistant Standalone chat UI + OpenAI proxy — 80+ models, branching, comparison, playground, caching, auth
Exploration Journal Narrative of reverse-engineering Google APIs — batchexecute, CDP, ARGUS, Chrome MCP
Integrations SDK Complete API reference — LMStudio, NotebookLM (56 methods), Copilot, Colab, Sheets
NotebookLM SDK Design SDK internals — 42 rpcids, 25 gRPC methods, payload formats, gotchas
NLM API Reference Auto-generated ARGUS API catalog for NotebookLM
Gemini API Reference Auto-generated ARGUS API catalog for Gemini
AI Studio API Reference Auto-generated ARGUS API catalog for AI Studio

Development

Doc Description
Contributing Scene creation, skill writing, code conventions, testing
Testing Smart test system, fixtures, pytest conventions
API Reference REST endpoints, Socket.IO events, all scene and service routes

Project History

Archived narrative docs are in docs/archive/ — see archive/README.md:

  • PROJECT_JOURNAL.md — Sprint history from v0.51b through v0.84b
  • PROJECT_JOURNAL_NLM.md — NotebookLM development journal
  • NLM_JOURNEY.md — Reverse engineering story
  • PROJECT_HINDSIGHT.md — v0.83b refactor retrospective
  • SYSTEM_AUDIT.md — Historical system audits
  • GAP_ANALYSIS.md — v1.26 gap analysis
  • AGENT_ONBOARDING.md — Copilot workflow (superseded by .github/agents/)
  • animation_creation.md — Penthouse animation guide