Release v0.0.17

.claude/.gitignore

@@ -0,0 +1,2 @@
+# MEOW-generated hooks - do not commit
+settings.json

.gitignore

@@ -57,6 +57,8 @@ logs/
 .coverage
 htmlcov/
 tests/output/
+playwright-report/
+test-results/
 
 # Frontend (Node.js/Vue)
 frontend/node_modules/
@@ -75,3 +77,4 @@ web/.meow/
 .browserflow/
 node_modules/
 packages/
+legacy/

README.md

@@ -54,6 +54,15 @@ New to ComfyGit? The setup wizard guides you through creating your first workspa
 
 <img width="1596" height="965" alt="Screenshot 2025-12-08 185458" src="https://github.com/user-attachments/assets/b46e00d6-e13c-4320-b4cc-a1d575e4f0f9" />
 
+### Keyboard Shortcuts
+
+| Shortcut | Action |
+|----------|--------|
+| `Alt+Shift+L` | Open ComfyGit Panel |
+| `Alt+Shift+K` | Quick Commit |
+
+Customize these in ComfyUI's **Settings → Keyboard Shortcuts**.
+
 ---
 
 ## Getting Started

docs/huggingface-hub-guide.md

@@ -0,0 +1,360 @@
+# HuggingFace Hub Python Library - Quick Reference
+
+Concise guide for finding, downloading, and authenticating with HuggingFace Hub.
+
+## Installation
+
+```bash
+pip install huggingface_hub
+# or
+uv add huggingface_hub
+```
+
+## Authentication
+
+### Methods (choose one)
+
+```python
+# 1. Environment variable (recommended for scripts)
+# Set HF_TOKEN=hf_xxx in environment
+
+# 2. Login once (persists token to ~/.cache/huggingface/token)
+from huggingface_hub import login
+login(token="hf_xxx")  # or login() for interactive prompt
+
+# 3. Pass token directly to functions
+hf_hub_download(..., token="hf_xxx")
+
+# 4. HfApi client with token
+from huggingface_hub import HfApi
+api = HfApi(token="hf_xxx")
+```
+
+### Token precedence
+1. Explicit `token=` parameter
+2. `HF_TOKEN` environment variable
+3. Stored token at `~/.cache/huggingface/token`
+
+### Other auth functions
+```python
+from huggingface_hub import logout, auth_switch, auth_list
+logout()                    # Remove stored token
+auth_list()                 # List stored tokens
+auth_switch("token_name")   # Switch between tokens
+```
+
+## Finding Models
+
+### List models with filters
+
+```python
+from huggingface_hub import list_models, HfApi
+
+# Basic listing
+models = list_models()
+
+# With filters
+models = list_models(
+    filter="text-generation",           # Task/tag filter
+    author="meta-llama",                # Filter by author
+    search="llama",                     # Search in name
+    library_name="transformers",        # Filter by library
+    pipeline_tag="text-generation",     # Filter by pipeline
+    sort="downloads",                   # Sort: downloads, likes, trending_score, created_at
+    limit=10,                           # Max results
+    token="hf_xxx",                     # For private models
+)
+
+# Iterate results
+for model in models:
+    print(f"{model.id}: {model.downloads} downloads")
+```
+
+### Get model info
+
+```python
+from huggingface_hub import model_info
+
+info = model_info("meta-llama/Llama-2-7b", token="hf_xxx")
+print(info.id, info.sha, info.downloads, info.pipeline_tag)
+print(info.siblings)  # List of files in repo
+```
+
+### Check if model/file exists
+
+```python
+from huggingface_hub import repo_exists, file_exists
+
+repo_exists("meta-llama/Llama-2-7b")                    # True/False
+file_exists("meta-llama/Llama-2-7b", "config.json")     # True/False
+```
+
+### List files in repo
+
+```python
+from huggingface_hub import list_repo_files
+
+files = list_repo_files("meta-llama/Llama-2-7b", token="hf_xxx")
+```
+
+## Downloading
+
+### Download single file
+
+```python
+from huggingface_hub import hf_hub_download
+
+# To cache (default: ~/.cache/huggingface/hub)
+path = hf_hub_download(
+    repo_id="meta-llama/Llama-2-7b",
+    filename="config.json",
+    token="hf_xxx",
+)
+
+# To specific folder (replicates repo structure)
+path = hf_hub_download(
+    repo_id="meta-llama/Llama-2-7b",
+    filename="config.json",
+    local_dir="/path/to/models/llama2",  # Key param!
+    token="hf_xxx",
+)
+
+# File in subfolder
+path = hf_hub_download(
+    repo_id="stabilityai/stable-diffusion-xl-base-1.0",
+    filename="model_index.json",
+    subfolder="scheduler",
+    local_dir="/models/sdxl",
+)
+
+# Specific revision (branch, tag, or commit hash)
+path = hf_hub_download(
+    repo_id="meta-llama/Llama-2-7b",
+    filename="config.json",
+    revision="main",  # or "v1.0" or commit hash
+)
+
+# Force re-download
+path = hf_hub_download(..., force_download=True)
+
+# Use only cached files (offline mode)
+path = hf_hub_download(..., local_files_only=True)
+```
+
+### Download entire repo (snapshot)
+
+```python
+from huggingface_hub import snapshot_download
+
+# To cache
+path = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    token="hf_xxx",
+)
+
+# To specific folder
+path = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    local_dir="/path/to/models/llama2",  # Key param!
+    token="hf_xxx",
+)
+
+# Filter files with patterns (glob syntax)
+path = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    allow_patterns=["*.safetensors", "*.json"],  # Only these
+    ignore_patterns=["*.bin", "*.h5"],           # Skip these
+    local_dir="/models/llama2",
+)
+
+# Download specific revision
+path = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    revision="main",
+    local_dir="/models/llama2",
+)
+
+# Parallel downloads (default: 8 workers)
+path = snapshot_download(..., max_workers=16)
+```
+
+### Dry run (check what would download)
+
+```python
+# Single file
+info = hf_hub_download(
+    repo_id="meta-llama/Llama-2-7b",
+    filename="config.json",
+    dry_run=True,
+)
+print(info.file_size, info.is_cached, info.will_download)
+
+# Full repo
+infos = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    dry_run=True,
+)
+for info in infos:
+    print(f"{info.filename}: {info.file_size} bytes, cached={info.is_cached}")
+```
+
+## Download Locations
+
+### Key parameters
+
+| Parameter | Behavior |
+|-----------|----------|
+| `local_dir` | Download to this folder, replicating repo structure. Creates `.cache/huggingface/` inside for metadata. |
+| `cache_dir` | Use this as cache location instead of default. Files stored with content-addressed naming. |
+| (neither) | Uses default cache: `~/.cache/huggingface/hub` |
+
+### Environment variables
+
+```bash
+HF_HOME=~/.cache/huggingface           # Base dir for all HF data
+HF_HUB_CACHE=$HF_HOME/hub              # Where repos are cached
+HF_TOKEN=hf_xxx                        # Auth token
+HF_HUB_OFFLINE=1                       # Force offline mode
+HF_HUB_DISABLE_PROGRESS_BARS=1         # No progress bars
+```
+
+## Repo Types
+
+For datasets or spaces, add `repo_type`:
+
+```python
+# Download dataset
+snapshot_download(
+    repo_id="squad",
+    repo_type="dataset",
+    local_dir="/data/squad",
+)
+
+# Download space
+snapshot_download(
+    repo_id="gradio/hello_world",
+    repo_type="space",
+    local_dir="/spaces/hello",
+)
+
+# List datasets
+list_datasets(filter="text-classification")
+
+# Dataset info
+dataset_info("squad")
+```
+
+## HfApi Client
+
+For multiple operations, use the client:
+
+```python
+from huggingface_hub import HfApi
+
+api = HfApi(token="hf_xxx")
+
+# All methods available
+api.list_models(author="meta-llama")
+api.model_info("meta-llama/Llama-2-7b")
+api.hf_hub_download(...)
+api.snapshot_download(...)
+api.list_repo_files(...)
+api.repo_exists(...)
+api.file_exists(...)
+api.whoami()  # Get current user info
+```
+
+## Common Patterns
+
+### Download model to specific folder
+
+```python
+from huggingface_hub import snapshot_download
+import os
+
+def download_model(repo_id: str, dest_dir: str, token: str = None):
+    """Download a model to a specific directory."""
+    return snapshot_download(
+        repo_id=repo_id,
+        local_dir=dest_dir,
+        token=token or os.getenv("HF_TOKEN"),
+        allow_patterns=["*.safetensors", "*.json", "*.txt"],
+        ignore_patterns=["*.bin", "*.h5", "*.ckpt"],
+    )
+
+# Usage
+path = download_model("meta-llama/Llama-2-7b", "/models/llama2")
+```
+
+### Check size before downloading
+
+```python
+from huggingface_hub import snapshot_download
+
+infos = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    dry_run=True,
+    allow_patterns=["*.safetensors"],
+)
+total_size = sum(f.file_size for f in infos if f.will_download)
+print(f"Will download: {total_size / 1e9:.2f} GB")
+```
+
+### Download with progress callback
+
+```python
+from huggingface_hub import snapshot_download
+from tqdm import tqdm
+
+class CustomTqdm(tqdm):
+    def __init__(self, *args, **kwargs):
+        kwargs.setdefault('unit', 'B')
+        kwargs.setdefault('unit_scale', True)
+        super().__init__(*args, **kwargs)
+
+path = snapshot_download(
+    repo_id="meta-llama/Llama-2-7b",
+    local_dir="/models/llama2",
+    tqdm_class=CustomTqdm,
+)
+```
+
+## Error Handling
+
+```python
+from huggingface_hub import hf_hub_download
+from huggingface_hub.errors import (
+    RepositoryNotFoundError,
+    RevisionNotFoundError,
+    EntryNotFoundError,
+    GatedRepoError,
+)
+
+try:
+    path = hf_hub_download("meta-llama/Llama-2-7b", "config.json")
+except RepositoryNotFoundError:
+    print("Repo doesn't exist or is private")
+except GatedRepoError:
+    print("Need to accept license at huggingface.co")
+except RevisionNotFoundError:
+    print("Branch/tag/commit not found")
+except EntryNotFoundError:
+    print("File not found in repo")
+```
+
+## Gated Models
+
+Some models require accepting terms on the website first:
+
+1. Go to model page (e.g., https://huggingface.co/meta-llama/Llama-2-7b)
+2. Accept the license/terms
+3. Download with authenticated token
+
+```python
+# After accepting terms on website
+path = hf_hub_download(
+    repo_id="meta-llama/Llama-2-7b",
+    filename="config.json",
+    token="hf_xxx",  # Required for gated models
+)
+```