AI Girlfriend

Language / 语言 / 言語： 🇨🇳 中文 · 🇬🇧 English · 🇯🇵 日本語

100% Local · Fully Private · Zero API Dependencies

All conversations, voice, images, and character animations are generated on your own machine. No cloud servers, no third-party APIs, no risk of data leakage. Your AI girlfriend belongs to you, and only you.

🗳️ Fourth girlfriend voting in progress - please vote on Issues · Config tutorial: BV16XTV6fEoH · qq: 580322386 ⚠️ Default scripts are for NVIDIA GPUs. AMD GPU users: see the AMD_GPU/ folder.

An uncensored AI girlfriend harem project powered by OpenClaw + QQ Bot + Telegram Bot + llama.cpp + GPT-SoVITS + ComfyUI + Sakura Desktop Pet + Live2D - running entirely on your own machine.

Characters: Supports hot-swappable AI girlfriends with isolated memories per character.

From Starry Moonlit Café & the Butterfly of Death. Tall, aloof, cool exterior with a hidden warmth. A natural quietly-dominant type - she takes the lead, teases you gently, and guards you fiercely. Speaks little, but every word hits.

From ATRI -My Dear Moments-. Petite, innocent, endlessly curious - a bright-eyed girl who wears her heart on her sleeve. Runs toward the future with a smile, dragging you along. The polar opposite of Natsume: bubbly and expressive where Natsume is reserved, emotionally transparent where Natsume is guarded, playful where Natsume is composed. If Natsume is the cool winter night, ATRI is the warm summer sun.

From Dimension W Lovers!!. Former student council president and the academy's strongest anti-kaiju combatant. Silver-white hair with pink tips, pale blue eyes - cool-headed, restrained, and fiercely responsible. She's not good at smooth words or easy smiles; her care is direct and clumsy, like a command: rest, eat, don't push yourself. In desktop pet form, she's learning that she doesn't have to bear everything alone - that protecting someone's ordinary everyday life from this side of the screen is enough. A quiet guardian: silent but watchful, loyal but stubborn, the senpai who stays by your side without being asked.

⚠️ First step: Run quick_setup.ps1 to configure paths and language.

This wizard will:

1. Let you choose the default Agent language (Chinese / Japanese / English) - copies the corresponding AGENTS_*.md to DEFAULT_AGENT.md
2. Auto-detect your installed tools (ComfyUI, GPT-SoVITS, llama.cpp, embedding models)
3. Prompt you for any paths it can't find
4. Generate config.yaml with all paths, ready for download-models.ps1

powershell -ExecutionPolicy Bypass -File quick_setup.ps1

Or use the new auto-download script (clones ComfyUI and GPT-SoVITS from git automatically):

# Windows
powershell -ExecutionPolicy Bypass -File setup-deps.ps1

# Linux / macOS
bash setup-dependencies.sh

This script:

• Detects your OS and GPU environment
• Clones ComfyUI from https://github.com/comfyanonymous/ComfyUI
• Clones GPT-SoVITS from https://github.com/RVC-Boss/GPT-SoVITS
• Updates config.yaml with the new paths

After setup completes, proceed with download-models.ps1 → setup-llama.ps1 → start.ps1.

👆 QQ Bot: text chat + TTS voice + ComfyUI image generation + character memory

👆 Shiki Natsume Live2D: real-time character animation with emotion-driven motions, lip-sync, and speech bubbles. Controlled via local HTTP bridge.

Personality opposite of Natsume, hot-swappable with isolated memory.

👆 ATRI Live2D: silver hair, ruby-red eyes, barefoot in a white dress - innocent and expressive.

👆 ATRI ComfyUI: AI image generation - seaside sunset, flowing white dress, warm golden-hour lighting.

Cool-headed guardian senpai, student council president and academy's strongest combatant - now your desktop companion.

👆 Yono Sakura Desktop Pet: silver-pink gradient hair, pale blue eyes, school uniform - reactive portrait expressions, proactive care reminders, and real-time TTS voice via GPT-SoVITS.

👆 Web Chat: browser-based chat interface at http://127.0.0.1:19270 - an alternative to QQ/Telegram bots. Connects directly to local daemon proxy → llama.cpp server. 8 GB VRAM can run fully without stopping.

🔊 Listen (click to play, ATRI Japanese):

🎧 tts_atori.mp3 (46KB, plays in browser)

👆 Artemis Studio - ComfyUI Workshop: Visual AI image generation console - freely choose character/outfit/scene/art style, one-click generation. Runs in parallel with llama (12GB+ VRAM).

🚀 Not a "future" plan anymore — it works today: Qwen-Drive-1.0-4B

📖 Full design: imagination.md | Bridge ref: skills/cosmos/BRIDGE_REFERENCE.md

NVIDIA Cosmos (community FP8 quant archived at skills/cosmos/) is a World Foundation Model that generates physics-consistent scene videos and understands spatial relationships.

The four core capabilities (LLM + TTS + ComfyUI + Live2D) are currently disconnected - the LLM doesn't know what Live2D is doing, ComfyUI doesn't sense conversational emotion. Cosmos fills the physical common-sense layer:

Qwen3.6-35B (Language Mind) ←→ Cosmos 3 Nano / Qwen-Drive-1.0 (Physical Mind)
   Language + Emotion             Spatial + Scene Generation

• ✅ Repo archived at skills/cosmos/
• ✅ Bridge design imagination.md + cosmos_check.py ready
• ✅ Qwen ↔ Cosmos dual-mind architecture designed
• ✅ 2027-01: RTX 5070 Ti Super (24 GB) — already running

• 🔄 Multi-Character Hot-Swap - One-click switch between AI girlfriends (Natsume ⇄ ATRI ⇄ Sakura); SOUL/IDENTITY/TTS weights/Live2D model all switch automatically, memories isolated per character
• 🃏 SillyTavern Character Card Import - Auto-detect and import PNG/JSON character cards; agent auto-switches persona on import
• 💬 Chat Log Import - Import SillyTavern JSONL conversation logs into memory/role_play/<character>/; agent restores context on role switch
• 🎤 TTS Voice Synthesis - Local GPT-SoVITS inference, Japanese voice (emotion-matched per dialogue), 3 character voice models (Natsume / ATRI / Sakura)
• 🎤 ASR Speech Recognition - Local Faster-Whisper small model (~1.5GB VRAM), coexists with llama; 99-language support
• 🎨 AI Image Generation - Local ComfyUI inference, SDXL/Illustrious models, 3 character prompt templates
• 🖥️ Sakura Desktop Pet - PySide6 desktop companion with proactive care, screen observation & local LLM awareness; supports 3 characters
• 🎭 Live2D Character Model - Real-time Live2D rendering with emotion-driven expressions & speech bubbles (Natsume / ATRI L2D; Sakura portrait mode)
• 🧠 Smart VRAM Tiering - Auto-detects GPU VRAM and picks the right strategy: ≥12GB keeps everything online (llama + skills); 8GB hot-swaps llama for GPU-heavy tasks; <8GB safe mode. Zero manual config
• 🎛️ Artemis Studio Console - Visual TTS + ComfyUI workshop, DIY voice & images anytime regardless of llama status - a true offline creative suite
• 💾 Roleplay Memory - Daily conversation summaries per character in memory/role_play/
• 🧠 Long-term Memory System - Powered by headroom (SmartCrusher + CCR) and mem0 (Qdrant vector database):
  • Chinese Embedding Boost - Added BGE-small-zh-v1.5 alongside all-MiniLM-L6-v2 for more accurate CN/JP/EN hybrid memory retrieval
  • SmartCrusher Context Trimming - Hard-caps chat history at 24 messages / 40K characters per LLM request
  • CCR (Curate-Consolidate-Retrieve) - Background worker extracts durable facts every 8 turns, writes to mem0 Qdrant
  • Vector + BM25 Hybrid Search - Semantic similarity + keyword matching via Qdrant + dual embedding models
  • Auto-Sync Bridge - Cron job syncs Qdrant → _mem0_auto.md every 30 min, making vector memories searchable by OpenClaw's native memory_search
  • Per-Character Isolation - user_id scoping in Qdrant; 4 independent memory spaces (sakura / natsume / enola / atori)
  • Recall Priority - Vector long-term memories > handwritten daily notes > SOUL base persona

See skills/behavior-engine/README.md and AGENTS_roleplay_EN.md#behavior-engine.

A layered decision engine ported from the sister-project girl-agent, giving each character an independent relationship score, conflict state, relationship stage, and hormonal cycle that drive their behavior and reply style.

Core loop: each turn produces a moodDelta (interest/trust/attraction/annoyance/cringe) → accumulated into the score → triggers conflict escalation/cool-down → auto-checks relationship stage transitions → shapes the LLM's reply style.

9 relationship stages: first meet → cold period → warming up → convinced → first date → early dating → stable dating → long-term → dumped

4-level conflict system: level 0 normal → level 1 slight sulk → level 2 in a huff → level 3 severe cold war → level 4 blocked/deleted

Hormonal cycle: a Gaussian cycle model simulates periodic swings in energy, irritability, affection, and libido, influencing reply length and tone.

State file: memory/role_play/<char>/relationship.json (independent per character, hot-loaded) Module location: skills/behavior-engine/

All models hosted on HuggingFace: TAOTAO777/ai-girlfriend-natsume

See models.yaml for full details.

📁 Embedding models path: embedding/all-MiniLM-L6-v2/ + embedding/bge-small-zh-v1.5/ (HF repo)

# Install huggingface-cli: pip install huggingface_hub
huggingface-cli login

# Download all models
huggingface-cli download TAOTAO777/ai-girlfriend-natsume --local-dir ./models

# Or download individual components:
huggingface-cli download TAOTAO777/ai-girlfriend-natsume llm/ --local-dir ./models
huggingface-cli download TAOTAO777/ai-girlfriend-natsume comfyui/ --local-dir ./comfyui
huggingface-cli download TAOTAO777/ai-girlfriend-natsume gpt-sovits-weights/ --local-dir ./gpt-sovits-weights
huggingface-cli download TAOTAO777/ai-girlfriend-natsume live2d-model/ --local-dir ./live2d-model

# Ternary-Bonsai-2-27B PTQ1_0 (mirror of our repo's llm/ folder, fits 8 GB VRAM):
huggingface-cli download TAOTAO777/ai-girlfriend-natsume llm/Ternary-Bonsai-2-27B-PTQ1_0.gguf --local-dir ./models

🇨🇳 Users in China: use hf-mirror.com - no VPN needed: set HF_ENDPOINT=https://hf-mirror.com then run hf download as usual.

1. Run quick_setup.ps1 - interactive wizard that generates config.yaml with your local paths
2. (Alternative) Copy config.example.yaml → config.yaml and edit manually
3. Place downloaded model files according to models.yaml, then update config.yaml paths

All Python/PS scripts read paths from config.yaml - no hardcoded paths to edit.

⚠️ Disclaimer: All models are community open-source. This project only provides mirror distribution, non-profit. Copyright belongs to original authors.

Running Qwen3.6-35B-A3B Genesis Hermes V13 MTP APEX Compact (MoE, 16.11 GiB, 34.66B params, 8/256 experts) via llama.cpp with speculative MTP (Multi-Token Prediction) decoding.

🚀 You don't hand-type llama-server args anymore. Every launch entry point (start.ps1, shiki_daemon.py, restart_llama_degraded.ps1) reads the launch parameters from config.yaml via skills/shared/llama_config.py, which auto-matches the active model filename against model_profiles and builds the full llama-server command. Models are auto-detected and parameters separate by profile — nothing is hardcoded.

See "Switching models" below for the one-liner.

⚙️ The baked-in launch commands are semi-hardcoded — treat them as a starting point, not gospel. The profile parameters in config.yaml / llama_config.py were tuned for the reference machine. Before trusting them on your own hardware, read LLAMA_TUNING.md (handwritten field notes: when to use -ngl 99 vs partial -ngl N vs --cpu-moe, MTP draft tuning, KV cache sizing, batch/ubatch, threads, context window) and decide the final llama-server command based on that guide plus your machine's GPU/RAM/CPU configuration. In short: pick the offload tier that matches your VRAM vs model size (partial -ngl N is fine for dense models that don't fully fit — it's a static split, not dynamic swapping), tune --spec-draft-n-max × --spec-draft-p-min until acceptance looks good, and size context/KV cache to your RAM. See the Qwen3.8-27B (Dense, Tooling Model) section below for the reference flags and live metrics.

The project root ships a fixed Jinja chat template (froggeric/Qwen-Fixed-Chat-Templates, pinned at v22.3 in chat_template.jinja) that overrides the template baked into the GGUFs. The official Qwen 3.5/3.6/3.8 templates contain engine restrictions, Python-specific Jinja logic, and regressions that break local inference and agent workflows — the most visible one is overthinking: the official Qwen 3.8 template hardcodes xhigh reasoning depth by default, which can exhaust the token budget on thinking before the model ever answers.

One file covers all Qwen 3.5 / 3.6 / 3.8 sizes, so it works unchanged for both local models. Launch plumbing: config.yaml → llama_chat_template: chat_template.jinja (relative to the project root), and llama_config.py resolves it to --chat-template-file — nothing hardcoded.

llama-server.exe ... --jinja --reasoning-preserve \
  --chat-template-file "D:\AI_Girlfriend\chat_template.jinja"

Switch between the two active models with a one-liner — the script kills the current llama-server, rewrites config.yaml (llama_model / llama_model_name / llama_model_id), re-resolves the profile, restarts, and waits for /health:

cd D:\AI_Girlfriend
# 27B dense (Qwen3.8-27B) — primary tooling model
.\skills\shared\restart_llama_degraded.ps1 -SwitchTo qwen3.8-27b

# 35B MoE (Hermes Genesis V13) — primary roleplay model
.\skills\shared\restart_llama_degraded.ps1 -SwitchTo qwen3.6-35b

-SwitchTo accepts the key in config.yaml → llama_model_map (e.g. qwen3.8-27b / qwen3.6-35b), or a substring (e.g. -SwitchTo 27b). Use -ForceBatch 1024 to lower batch size if you hit VRAM limits.

See LLAMA_TUNING.md for more llama tuning details.

Served on http://127.0.0.1:8080. Note: every argument pair in the PowerShell array must be comma-separated — a missing comma silently glues two tokens together.

💡 rea not specified — defaults to medium reasoning depth via the chat template (no injected thinking tokens, preserving KV-cache parity). This is the optimal setting for tooling/agent tasks where you want fast, direct responses.

Ternary-Bonsai-2-27B PTQ1_0 — hosted in this project's HF repo: TAOTAO777/ai-girlfriend-natsume → llm/Ternary-Bonsai-2-27B-PTQ1_0.gguf (same llm/ folder as the other two LLM models). Original source: base model prism-ml/Ternary-Bonsai-2-27B-gguf; PTQ1_0 ternary quant build by BoldingBuilds.

5.9 GB on disk (~5.5 GiB). Verified running on the reference 8 GB VRAM laptop with -ngl 99: the whole model fits on the card — no RAM layer split needed. Measured: decode ~35 t/s+, prefill ~300 t/s. Tuning notes: LLAMA_TUNING.md.

🔴 Two NON-NEGOTIABLE constraints for 8 GB VRAM:

1. KV cache MUST be Q4: -ctk q4_0 -ctv q4_0. Anything higher (f16 / f32 KV) will blow the VRAM budget immediately.
2. Context window MUST be -c ≤ 75000. With Q4 KV, weights (~5.5 GiB) + KV cache + compute buffers stay inside 8 GB only up to ~75K tokens. Anything larger does not fit on an 8 GB card.

Silicon Rider Bench is an agent benchmark that simulates a food-delivery rider working a virtual city: navigate, accept orders, pick up food, deliver on time, and manage battery — scoring total profit over a simulated 24-hour day. Same seed (622539) used across all runs for apples-to-apples comparison.

Models under test (all --seed 622539):

• deepseek-v4-flash (0731) — remote, unlimited-context baseline. Cloud-class agent ability (~Claude 4.6–4.8 tier in this benchmark).
• Hermes3.6-35B-A3B-Uncensored-Genesis-V9-MTP-APEX-Compact.gguf (current) — RTX 5070 Laptop, 8 GB VRAM, 32 GB DDR5 RAM

• Context length is the #1 lever: raising CONTEXT_HISTORY_LIMIT 25 → 100 lifted on-time rate 56.7% → 75% and slashed overtime penalty ¥107.9 → ¥44.7, pushing profit ¥411 → ¥525 (the model finally retains order deadlines + routes across turns).
• Local 35B MoE ≈ 85% of cloud flash: at 100 ctx the local quantized 35B hits ¥524.6 = 84.6% of dsv4-flash's ¥619.6, with on-time rate (75% vs 81.8%) and per-order profit (¥18.74 vs ¥18.77) essentially tied.
• 6× cheaper: flash burned 24.39M tokens (unlimited ctx); local 100-ctx used only 4.08M for 5/6 of the profit → 5× better token efficiency, at zero API cost.
• Remaining gap: route efficiency (1.68 vs 1.34) — the 35B-A3B's 3B active params still underperform flash on multi-leg optimal route planning.

Verdict: after quantization fine-tuning, the Hermes3.6-tuned Qwen3.6 35B's agentic ability is essentially on par with Claude Opus 4.6!

🧪 Full logs & reports in docs/silicon-rider-bench-622539/ (COMPARISON-622539.md + per-run summaries).

Qwen3.6 MoE uses SSM (Gated Delta Net) hybrid attention with --kv-unified.

⚠️ Known Limitation: Cross-turn prompt cache reuse is not supported (SSM architecture limitation). Each request triggers full context re-processing. Longer conversations = higher first-token latency (~55s for 59k tokens).

Mitigations:

• Periodic /reset (Natsume writes roleplay summaries to memory/role_play/ before resetting)
• Restore context from summaries on startup, keeping actual token count in 5K-20K range
• config-patch.json sets OpenClaw contextWindow to 262144 to match model capacity

Primary tooling/assistant dense model. Runs via llama.cpp with built-in MTP speculative decoding (no separate draft GGUF needed). Auto-detected via config.yaml → model_profiles (qwen3.8-27b-mtp).

Switch to it, or launch manually:

# Preferred: auto-switch + auto-params (see "Switching models" above)
.\skills\shared\restart_llama_degraded.ps1 -SwitchTo qwen3.8-27b

Reference hardware: i9-14900HX + RTX 5070 Laptop (8 GB) + 64 GB RAM. The model weights are split via a partial offload -ngl 14 (first 14 layers on GPU, rest in RAM with --no-mmap); KV cache uses --cache-ram 2000; the MTP draft context is offloaded fully to the GPU (--spec-draft-ngl 99) so speculative decoding stays fast on an 8 GB card. Q4_K_S quantization keeps the model at ~15.8 GB — the sweet spot for dense 27B on consumer hardware. rea not specified — defaults to medium reasoning depth via the chat template. Flag-by-flag notes, key parameters, and live-log metrics follow below.

💡 27B dense on 8 GB VRAM — key parameters explained:

• -ngl 14 — 14 layers offloaded to GPU (static split; rest in system RAM). For an 8 GB card with a ~15.8 GB Q4_K_S model, this is the sweet spot that fits without OOM while still getting meaningful GPU acceleration. Adjust up/down based on your actual VRAM.
• -ctk q4_0 -ctv q4_0 — KV cache quantized to 4-bit, halving VRAM usage for the context window. Essential for large context with limited VRAM.
• --cache-ram 2000 — 2 GB RAM budget for the KV cache on the CPU side.
• -c 100000 — 100K token context window (the model's effective limit at this quantization).
• --spec-draft-n-max 3 — MTP speculative decoding drafts up to 3 tokens ahead; Qwen3.8 ships its own MTP head.
• --spec-draft-p-min 0.88 — Only accept draft tokens with ≥88% confidence, keeping the acceptance rate high.
• --spec-draft-ngl 99 — Offload the entire draft context to GPU for faster speculative decoding.
• Quantization: Q4_K_S — ~15.8 GB model size, excellent quality/VRAM balance for dense 27B on consumer hardware. This is a dense (non-MoE) model, so all 27B parameters are active at inference (vs MoE which activates a subset).
• rea not specified — defaults to medium reasoning depth via the chat template (no injected thinking tokens, preserving KV-cache parity).

Flag notes (dense 27B profile, 8 GB VRAM optimum, Q4_K_S):

📈 MTP explained: with --spec-draft-n-max 5 + --spec-draft-p-min 0.84, llama.cpp asks the MTP head to propose up to 5 next tokens, then keeps each only if its probability is ≥0.84. In practice ~90–100% of drafted tokens are accepted (mean accepted run length ≈ 3.2–5.3), so effective throughput is roughly 3–5× a single speculative token per forward pass while the 8 GB card stays within its VRAM cap.

💡 MoE vs Dense: The 35B MoE activates only ~3B parameters per token (8/256 experts) and fits GPU well (48 tok/s). The 27B dense activates all 27B, exceeding 8 GB VRAM, so it splits to CPU/RAM via -ngl 14 and decodes ~4–5 tok/s with MTP. Use the 27B dense when you want full 27B activation for tooling / agent tasks; use the 35B MoE for fast roleplay. The Q4_K_S quant (~15.8 GB) is the sweet spot for dense 27B on consumer hardware — excellent quality while fitting on 8 GB VRAM with partial offload.

The system auto-detects GPU VRAM and selects the optimal run mode, no manual config:

┌────────────────────────────────────┬────────────┬────────────┬────────────┬────────────┐
│ VRAM Tier                          │ TTS        │ ComfyUI    │ llama      │ ASR        │
├────────────────────────────────────┼────────────┼────────────┼────────────┼────────────┤
│ Tier 0: <8GB                       │ Stop llama │ Stop llama │ Killed     │ Killed     │
│ Tier 1: 8-12GB (current)           │ Stop llama │ Stop llama │ Killed     │ No kill    │
│ Tier 2: ≥12GB                      │ No kill    │ No kill    │ Always on  │ No kill    │
└────────────────────────────────────┴────────────┴────────────┴────────────┴────────────┘

Current setup (8GB VRAM):

8 GB Total VRAM
├── llama-server resident: ~5.8 GB (model 4.6G + KV cache 1.2G)
├── Free: ~2.2 GB
│
├── TTS inference: stop llama → ~8 GB free → resume llama (~70s)
├── ComfyUI generation: stop llama → ~8 GB free → resume llama (~120s)
├── Artemis Studio (TTS/ComfyUI workshop): standalone - works regardless of llama
└── ASR / Live2D / Embedding: always online, unaffected by VRAM tiering

<PROJECT_DIR>/                            # OpenClaw workspace root
├── start.ps1                             # 🚀 One-click launch: llama + headroom + Live2D + Gateway
├── artemis_headroom_proxy.py             # Headroom proxy (19251): mem0 injection + SmartCrusher + routing
├── shiki_daemon.py                       # Daemon (19260/19270): WebChat backend + auto-inject provider
├── quick_setup.ps1                       # 🛠 Interactive path config wizard
├── config.yaml                           # Generated config
├── download-models.ps1                   # One-click model download (Windows)
├── download-models.sh                    # One-click model download (Linux/macOS)
├── setup-llama.ps1                       # Auto-detect HW + configure llama.cpp (Win)
├── setup-llama.sh                        # Auto-detect HW + configure llama.cpp (Linux/macOS)
├── setup-openclaw.ps1                    # One-click OpenClaw install + deploy (Win)
├── setup-openclaw.sh                     # One-click OpenClaw install + deploy (Linux/macOS)
├── setup-all.ps1                         # 🚀 All-in-One mega script (Windows)
├── setup-all.sh                          # 🚀 All-in-One mega script (Linux/macOS)
├── config-qqbot.json                     # QQ Bot config patch
├── config-telegram.json                  # Telegram Bot config patch
├── config-patch.json                     # OpenClaw LLM config patch
├── AGENTS.md                             # Agent behavior rules
├── SOUL.md                               # Character personality
├── IDENTITY.md                           # Character identity
├── USER.md                               # User info
├── HEARTBEAT.md                          # Heartbeat config
├── TOOLS.md                              # Tool quick reference
├── models.yaml                           # Model catalog + download links
├── LLAMA_TUNING.md                       # ⚙️ Handwritten llama.cpp tuning field notes (read before trusting baked-in launch args)
├── imagination.md                        # 🔮 Cosmos WFM integration vision (future)
├── README.md                             # This file
├── .gitignore
├── live2d/                               # Live2D character model (Cubism 4 Core)
│   ├── index.html                        # Default (Shiki Natsume)
│   ├── index_atri.html                   # ATRI variant
│   ├── index_upper.html                  # Natsume upper-body variant
│   ├── index_atri_upper.html             # ATRI upper-body variant
│   ├── live2dcubismcore.min.js           # Cubism Core 4 (207 KB)
│   ├── plid-v5-bundle.js                 # pixi-live2d-display v0.5.0 bundle
│   ├── live2d-bridge.mjs                 # HTTP (19200) + WebSocket (19201) bridge
│   ├── switch_model.ps1                  # Model switcher (natsume / atri)
│   ├── pixi.min.js, pixi-shim.js         # PIXI.js v7 rendering
│   ├── model/shiki_natsume/              # Natsume model (14 textures, 42 motions, 41 sounds)
│   └── model/atri/                       # ATRI model (2 textures, 620 voice mp3, 8 motions)
├── ren_pro_jp/                           # Ren'Py dialog engine (planned)
├── memory/                               # [.gitignore] Runtime memory
│   └── role_play/                        # Roleplay conversation logs
├── media/                                # [.gitignore] Generated media
│   ├── audio/                            # TTS voice output
│   ├── images/                           # ComfyUI image output
│   └── *.gif                             # README demo GIFs
├── docs/
│   ├── telegram-setup.md                 # Telegram Bot setup guide
│   └── qqbot-setup.md                    # QQ Bot setup guide
└── skills/
    ├── live2d/                           # Live2D control skill
    │   ├── SKILL.md                      # Motion/expression reference + API guide
    │   ├── scripts/start-live2d.ps1      # Live2D launcher
    │   └── media/                        # Shared media output
    ├── tts/
    │   ├── SKILL.md                      # TTS invocation guide
    │   ├── run_tts.ps1                   # TTS launcher script
    │   ├── tts_call.py                   # GPT-SoVITS inference
    │   └── ref_wavs/                     # Reference audio clips
    ├── comfyui/
    │   ├── SKILL.md                      # ComfyUI invocation guide
    │   ├── run_comfyui.ps1               # ComfyUI launcher script
    │   ├── comfyui_call.py               # ComfyUI inference
    │   ├── prompt_template.md            # Character prompt template
    │   └── custom_prompt.txt             # Custom extra prompt
    ├── asr/                              # Speech recognition skill
    │   ├── run_asr.ps1                   # Faster-Whisper launcher (~1.5GB VRAM)
    │   └── asr_call.py                   # Whisper small model inference
    ├── shared/                           # Shared infrastructure
    │   ├── embedding_server.py           # OpenAI-compatible embedding API (9999, dual model)
    │   ├── mem0_bridge.py                # mem0 Qdrant → OpenClaw memory bridge
    │   ├── start_embedding_server.ps1    # Auto-start embedding server
    │   ├── vram.py                       # VRAM tier auto-detection
    │   ├── VRAM_LEVELS.md                # VRAM tier documentation
    │   ├── llama_lifecycle.py            # Llama start/stop management
    │   └── llama_utils.py                # Llama utility functions
    ├── sakura/                           # Sakura Desktop Pet (PySide6 GUI)
    │   ├── SKILL.md                      # Sakura skill documentation
    │   ├── main.py                       # Application entry point
    │   ├── install.bat                   # Windows dependency installer
    │   ├── start.bat                     # Windows launcher
    │   └── app/                          # Source code
    ├── cosmos/                           # 🔮 NVIDIA Cosmos WFM (future hardware)
    │   ├── BRIDGE_REFERENCE.md           # Cosmos ↔ AI Girlfriend bridge design
    │   ├── cosmos_check.py               # Hardware VRAM detection script
    │   ├── cookbooks/                    # Official tutorial examples
    │   └── README.md                     # Upstream documentation
    ├── llama-management.md               # VRAM management architecture doc
    ├── llama-watchdog.ps1                # Llama health check
    ├── cleanup_orphans.ps1               # Orphan process cleanup
    ├── behavior-engine/                  # 💖 Relationship system (behavior engine)
    │   ├── engine.py                     # State load/save/update/reset
    │   ├── hormones.py                   # Hormonal cycle (Gaussian model)
    │   ├── conflict.py                   # 4-level conflict system
    │   ├── stages.py                     # 9 relationship stages
    │   ├── behavior_tick.py              # Behavior decision layer
    │   ├── online_tick.py                # Online/sleep simulation
    │   ├── daily_life.py                 # Daily schedule
    │   ├── README.md                     # Design doc
    │   └── SKILL.md                      # Usage guide
    └── character_importer/               # SillyTavern character card auto-import

Artemis now supports Claude Code as a parallel agent runtime alongside OpenClaw. Claude Code connects via MCP to access all Artemis capabilities - with a built-in AgentRQ-compatible task queue for human-agent collaboration.

┌─────────────────────────────────────────────────────────┐
│  Task Board (http://127.0.0.1:19280)                    │
│  Create task → assignee: agent → notstarted             │
└───────────────────────┬─────────────────────────────────┘
                        │ SQLite (.claude/task_queue.db)
                        ▼
┌─────────────────────────────────────────────────────────┐
│  Claude Code (terminal)                                 │
│  CLAUDE.md → getNextTask() → ongoing → execute          │
│  Artemis tools → TTS / ComfyUI / Live2D / memory        │
│  reply() → updateTaskStatus(completed)                  │
└─────────────────────────────────────────────────────────┘

Claude Code automatically runs a task loop on startup:

1. getWorkspace() - check workspace status
2. getNextTask() - dequeue next pending task
3. updateTaskStatus(taskId, "ongoing") - claim it
4. Execute using Artemis tools (TTS, ComfyUI, etc.)
5. reply(taskId, "Done!") - report result
6. updateTaskStatus(taskId, "completed") - mark done
7. Loop back to getNextTask()

# Prerequisites: npm install -g @anthropic-ai/claude-code
# Start Shiki Daemon first (.\shiki.cmd), then:

# Full AgentRQ workflow (Task Board + Claude Code)
.\claude-code.ps1

# Task Board only (browser UI, no Claude)
.\claude-code.ps1 -BoardOnly

# Stop the task board
.\claude-code.ps1 -KillBoard

Then open http://127.0.0.1:19280 - create tasks, watch Claude Code pick them up.

✨ TTS, ComfyUI, and Live2D are fully self-contained. No external downloads at runtime - all model weights (skills/sovits/, skills/comfyui_core/), Python scripts, JS libraries (live2d/pixi.min.js, live2d/plid-v5-bundle.js), and Cubism Core 4 (live2d/live2dcubismcore.min.js) are bundled locally.

🧠 Headroom token-saving - skills/headroom/ (SmartCrusher + ContentRouter + CCR). Compress large tool outputs in dev scenarios before they hit the context window. See AGENTS.md for API usage.

One command, from scratch to a fully functional AI girlfriend:

Windows:

powershell -File setup-all.ps1

Linux / macOS:

bash setup-all.sh

Automated pipeline: environment check → model download → llama.cpp setup → OpenClaw install → Sakura desktop pet → workspace deploy → path check → launch → verify.

Supports resume from breakpoint. Flags: --skip-model-download, --skip-llama-setup, --skip-openclaw-setup, --skip-sakura-setup, --dry-run, --no-start

Install OpenClaw Gateway and deploy the AI Girlfriend workspace:

Windows:

powershell -File setup-openclaw.ps1

Linux / macOS:

bash setup-openclaw.sh

This script installs Node.js, OpenClaw Gateway, deploys workspace files, installs daemon, and applies config patch.

Flags: --skip-node, --skip-deploy, --skip-daemon, --no-onboard

Windows:

pip install huggingface_hub
huggingface-cli login
powershell -File download-models.ps1

Linux / macOS:

pip install huggingface_hub
huggingface-cli login
bash download-models.sh

Downloads all 5 model files (~31.7 GB) from HuggingFace with progress reporting and resume support.

Auto-detects GPU, VRAM, CPU cores, RAM and generates optimized launch configs.

Windows:

powershell -File setup-llama.ps1

Linux / macOS:

bash setup-llama.sh

API Key (optional, recommended):

From this version, llama-server enables API key authentication by default (for security and extensibility). Configure in config.yaml:

llama_api_key: "123456"   # change to your own key; leave empty to skip --api-key

• After setting, the llama-server inference endpoints (/v1/chat/completions, etc.) require requests with Authorization: Bearer <key> or api_key:<key>.
• /health remains unauthenticated (health checks unaffected).
• All clients (headroom proxy / sakura / shiki_daemon forwarding) will automatically read llama_api_key and include the key, no additional config needed.
• When connecting to CCR, the CCR provider config also needs the upstream API key set to the same value.

powershell -File quick_setup.ps1

Interactive wizard - enter your local paths once, all scripts are updated automatically.

# One-click start all services (llama + Embedding + Live2D + Gateway)
powershell -File start.ps1

Startup sequence:

[1/8] llama-server        (8080, Qwen3.6-35B-A3B-MTP, --no-mmap, --spec-type draft-mtp)
[2/8] Embedding Server    (9999, all-MiniLM + BGE dual models, CPU, ~100MB RAM)
[3/8] VRAM Tier Detection (auto-selects whether TTS/ComfyUI stops llama)
[4/8] Headroom Proxy      (19251, mem0 memory injection + SmartCrusher compression + cloud routing)
[5/8] Live2D Bridge       (19200, pixi-live2d-display)
[6/8] OpenClaw Gateway    (18789, auto-injects local-llama provider)
[7/8] llama-watchdog      (crash auto-restart)
[8/8] Web Chat Daemon     (19260 API + 19270 webchat, --no-llama)

Shutdown: shiki.cmd -Stop - gracefully stops all services (llama → live2d → sakura → embedding → comfyui → gateway → cleanup).

# Start the bridge
Start-Process node -ArgumentList "live2d-bridge.mjs" -WorkingDirectory live2d -WindowStyle Hidden

# Open in standalone window (Chrome app mode)
Start-Process chrome -ArgumentList "--new-window --app=http://localhost:19200/index.html --window-size=450,650"

Live2D runs in a frameless Chrome window - place it anywhere on your desktop.

# Llama health check (every 10 min)
schtasks /create /tn "llama-watchdog" `
  /tr "powershell -File C:\Users\<you>\.openclaw\workspace\skills\llama-watchdog.ps1" `
  /sc minute /mo 10

# Orphan process cleanup (hourly)
schtasks /create /tn "cleanup-orphans" `
  /tr "powershell -File C:\Users\<you>\.openclaw\workspace\skills\cleanup_orphans.ps1" `
  /sc hourly /mo 1

OpenClaw Gateway (18789)
  ├─ <provider>/<model-id>           → Direct to original backend (skips headroom)
  ├─ local-llama/llama-local          → 19251 → llama-server:8080
  └─ local-llama/<model-id>           → 19251 → original backend (via headroom+mem0)
         │
         ▼
  headroom proxy (19251)
    ├─ [1] mem0 character memory injection (Qdrant vector search)
    ├─ [2] SmartCrusher 5-dim compressed conversation history
    └─ [3] Route to real backend
         ├─ llama-local → llama-server:8080
         └─ Cloud models    → sidecar finds real baseUrl

Add-only, no-change principle: start.ps1 auto-scans ~/.openclaw/openclaw.json on startup, adds local-llama provider (copies existing cloud model), original providers left as-is. Original baseUrl stored in ~/.openclaw/headroom_routes.json sidecar file. Zero-config after clone.

Immutable capability instructions with per-character memory isolation:

Recall priority: Vector long-term memories > handwritten daily notes > SOUL base persona

A complete web-based AI girlfriend chat interface, served locally at http://127.0.0.1:19270 by the shiki daemon.

The WebChat talks directly to the shiki daemon (:19260) which proxies to llama-server or OpenAI-compatible APIs. Character-switching is instant - each tab loads its own SOUL.md + IDENTITY.md + USER.md as the system prompt.

VRAM Orchestration Flow:

1. On startup: auto-detect GPU VRAM → determine tier (Tier 0/1/2)
2. Main session receives user request → assembles command
3. sessions_spawn(mode="run") creates sub-session
4. Tier 0/1: stop_llama() frees VRAM → TTS/ComfyUI inference → start_llama() resumes
5. Tier 2 (≥12GB): direct inference, llama stays online
6. Artemis Studio, Live2D, Embedding stay active throughout - unaffected
7. Sub-session writes .task_flags → announces back to main session
8. Main session reads media files → sends via <qqmedia> / MEDIA:
9. Background: CCR runs every ~8 turns, extracting long-term memories to Qdrant
10. Cron job syncs Qdrant → _mem0_auto.md every 30 min for native memory_search
11. Headroom proxy (19251) transparently intercepts local-llama/* requests → injects mem0 → compresses context → routes to real backend

• chat_template.jinja must stay at the project root (D:\AI_Girlfriend\chat_template.jinja) and must not be gitignored. It is the fixed froggeric v22.3 template referenced by config.yaml → llama_chat_template and passed via --chat-template-file. Deleting it or letting it be gitignored breaks llama launch args (the model falls back to a broken default template). .gitignore already has !chat_template.jinja to keep it tracked.
• Llama-server is offline for ~60-120s during TTS/ComfyUI inference on 8GB VRAM (Tier 1) - conversation pauses, but Live2D + Artemis Studio keep running. On 12GB+ (Tier 2), no interruption at all
• Llama-server does not support cross-turn prompt cache reuse (SSM limitation) - use periodic /reset
• Live2D requires Cubism Core 4 (not 5 or 6) - pixi-live2d-display v0.5.0 is built for Cubism 4 Framework; Core 5+ causes clipping/layer failures. Core 4 is bundled in live2d/live2dcubismcore.min.js - no CDN needed.

• @Rvosy - Creator of Sakura Desktop Pet, authorized for inclusion (Issue #38)
• @guansss - Creator of pixi-live2d-display
• Live2D Inc. - Cubism SDK (non-commercial use)
• AgentRQ - Inspiration for the AgentRQ-compatible task queue and MCP tool interface design
• headroom - Inspiration for SmartCrusher context compression + CCR (Curate-Consolidate-Retrieve) memory pipeline
• mem0 - Inspiration for Qdrant vector memory architecture + hybrid search design
• NVIDIA Cosmos - World Foundation Model, community FP8 quant archived at skills/cosmos/