Skip to content

scripts

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

scripts

README.md

LibreChat YAML Configuration Manager

Automated tool for maintaining LibreChat YAML configurations with up-to-date model lists from 20+ AI providers.

Quick Start

Prerequisites: Python 3.8+ and pip

# Install dependencies
pip install -r requirements.txt

# For local use, configure API keys
cp .env.example .env
# Edit .env with your API keys

Usage

Interactive Mode (Manual Updates)

Run the main update script:

python update.py

This will:

  1. Prompt whether to update YAML formatting (should not be required)
  2. Prompt whether to update model lists
  3. Create backups before any modifications
  4. Process configuration files

Automated Mode (No Prompts)

For automated/scheduled runs without user interaction:

# Update all YAML files
python update.py --automated

Or using environment variables:

AUTOMATED_MODE=true python update.py

Dedicated Automated Script

For GitHub Actions and other CI/CD pipelines:

# Runs automated update with YAML validation
cd scripts
python automated_update.py

This script:

  • Fetches latest models from all providers
  • Updates all YAML files
  • Validates YAML syntax
  • Returns appropriate exit codes:
    • 0: Success
    • 1: Update failed
    • 2: YAML validation failed

Individual Scripts

You can also run individual scripts directly:

Convert YAML formatting:

python convert_yaml_style.py

Update model lists:

# Interactive mode
python update_models.py

GitHub Actions (Automated Daily Updates)

This repository includes automated daily model updates via GitHub Actions.

Setup Instructions

  1. Add Repository Secrets

    Go to your repository Settings → Secrets and variables → Actions, and add the following secrets:

    AI302_API_KEY
APIPIE_API_KEY
COHERE_API_KEY
DEEPSEEK_API_KEY
FIREWORKS_API_KEY
GLHF_API_KEY
GROQ_API_KEY
HUGGINGFACE_TOKEN
HYPERBOLIC_API_KEY
KLUSTER_API_KEY
MISTRAL_API_KEY
NANOGPT_API_KEY
NVIDIA_API_KEY
OPENROUTER_KEY
PERPLEXITY_API_KEY
SAMBANOVA_API_KEY
TOGETHERAI_API_KEY
UNIFY_API_KEY
XAI_API_KEY

    Optional for notifications:

    NOTIFICATION_WEBHOOK  # Slack/Discord webhook URL

    Optional for automatic deploy webhook:

    DEPLOY_WEBHOOK_URL  # URL to trigger redeployment after model updates

  2. Workflow Configuration

    The workflow is located at .github/workflows/update-models.yml

    Schedule: Runs every day at 00:00 UTC

    Manual Trigger: Can be triggered manually from GitHub Actions tab

  3. How It Works

    • On Success: Changes are committed directly to main branch with notification
    • On YAML Validation Failure: Creates a PR for manual review and notifies @Berry-13
    • On Script Failure: Sends failure notification
    • Deploy Webhook: After successful commit, triggers deploy webhook (if configured)

  4. Deploy Webhook

    After model updates are committed, the workflow can trigger a deploy webhook:

    • Configuration: Add DEPLOY_WEBHOOK_URL as a repository secret
    • Behavior: If not configured, the step logs a warning and continues without error
    • Payload: JSON with event, date, repository, and run_id fields
    • Use case: Trigger redeployment on any platform (Docker server, cloud provider, etc.)

  5. Customizing the Schedule

    Edit .github/workflows/update-models.yml:

    on:
  schedule:
    - cron: '0 0 * * *'

    Examples:

    • Weekly (Mondays): '0 0 * * 1'
    • Every 6 hours: '0 */6 * * *'
    • Monthly (1st of month): '0 0 1 * *'

Notifications

The workflow sends notifications on:

  • ✅ Success: Models updated and committed
  • ⚠️ YAML Validation Failed: PR created for review
  • ❌ Failure: Script encountered errors

Configure webhook URL in repository secrets as NOTIFICATION_WEBHOOK.

Supported formats:

  • Slack incoming webhooks
  • Discord webhooks
  • Any webhook accepting JSON with text field

Directory Structure

librechat-config-yaml/
├── requirements.txt     # Python dependencies
├── README.md            # This file
├── scripts/             # Helper scripts
│   ├── README.md              # Scripts documentation
│   ├── update.py              # Main script
│   ├── convert_yaml_style.py  # Convert YAML formatting
│   ├── update_models.py       # Update model lists
│   ├── requirements.txt       # Python dependencies
│   ├── .env.example           # Example environment file
│   ├── .env                   # Your API keys (create this)
│   ├── ai302.py               # Fetch models from 302.AI
│   ├── apipie.py              # Fetch models from APIpie
│   ├── cohere.py              # Fetch models from Cohere
│   ├── deepseek.py            # Fetch models from Deepseek
│   ├── fireworks.py           # Fetch models from Fireworks
│   ├── github.py              # Fetch models from Github Models
│   ├── glhf.py                # Fetch models from GLHF.chat
│   ├── groq.py                # Fetch models from Groq
│   ├── huggingface.py         # Fetch models from HuggingFace
│   ├── hyperbolic.py          # Fetch models from Hyperbolic
│   ├── kluster.py             # Fetch models from Kluster
│   ├── mistral.py             # Fetch models from Mistral
│   ├── nanogpt.py             # Fetch models from NanoGPT
│   ├── nvidia.py              # Fetch models from Nvidia
│   ├── openrouter.py          # Fetch models from OpenRouter
│   ├── perplexity.py          # Fetch models from Perplexity
│   ├── sambanova.py           # Fetch models from SambaNova
│   ├── togetherai.py          # Fetch models from Together.ai
│   ├── unify.py               # Fetch models from Unify
│   └── xai.py                 # Fetch models from XAI
└── *.yaml               # LibreChat configuration files

Backup and Safety

  • All scripts create .bak files before modifying any YAML files
  • Logs are written to:
    • convert_yaml.log for YAML style conversion
    • update_models.log for model updates

Error Handling

  • Failed operations are logged with detailed error messages
  • The scripts will continue processing remaining files if one fails
  • A summary is displayed after completion showing:
    • Successfully processed files
    • Failed operations
    • Model count updates

Supported Providers

The tool can update model lists from:

  • 302.AI
  • APIpie
  • Cohere
  • Deepseek
  • Fireworks
  • Github Models
  • GLHF.chat
  • Groq
  • HuggingFace
  • Hyperbolic
  • Kluster
  • Mistral
  • NanoGPT
  • Nvidia
  • OpenRouter
  • Perplexity
  • SambaNova
  • Together.ai
  • Unify
  • XAI

Provider Script Validation Status

The provider scripts have been analyzed against their official API documentation:

Status Count Providers
✅ Validated 11 Cohere, DeepSeek, Fireworks, GitHub, Groq, HuggingFace, Mistral, NVIDIA, OpenRouter, TogetherAI, xAI
⚠️ Issues Found 4 APIpie (minor), NanoGPT, Perplexity, SambaNova
❓ Cannot Verify 5 302.AI, GLHF, Hyperbolic, Kluster, Unify

Key Findings

  • Perplexity & SambaNova: Use web scraping instead of APIs (fragile, may break if page structure changes)
  • Unify: Uses /v0/ API version (potentially experimental)
  • 5 providers: Lack public API documentation for verification
  • Most scripts: Well-implemented with proper authentication and error handling

Detailed Status

✅ Validated Scripts

These scripts have been verified against official API documentation and use proper API endpoints:

Provider Script API Endpoint
Cohere cohere.py https://api.cohere.com/v1/models
DeepSeek deepseek.py https://api.deepseek.com/models
Fireworks fireworks.py https://api.fireworks.ai/inference/v1/models
GitHub github.py https://models.inference.ai.azure.com/models
Groq groq.py https://api.groq.com/openai/v1/models
HuggingFace huggingface.py https://huggingface.co/api/models
Mistral mistral.py https://api.mistral.ai/v1/models
NVIDIA nvidia.py https://integrate.api.nvidia.com/v1/models
OpenRouter openrouter.py https://openrouter.ai/api/v1/models
TogetherAI togetherai.py https://api.together.xyz/v1/models
xAI xai.py https://api.x.ai/v1/models

⚠️ Scripts with Issues

Provider Issue Impact
APIpie Minor: No official docs found, but uses standard OpenAI-compatible endpoint Low
NanoGPT Web scraping from nano-gpt.com/api May break if page changes
Perplexity Scrapes documentation page instead of API Fragile, may break
SambaNova Scrapes community docs instead of API Fragile, may break

❓ Cannot Verify

These providers lack public API documentation:

  • 302.AI - No public API docs available
  • GLHF - No public API docs available
  • Hyperbolic - No public API docs available
  • Kluster - No public API docs available
  • Unify - Uses /v0/ endpoint (experimental)

Testing Limitations

API Key Requirements

Most provider scripts require valid API keys for testing:

Provider Key Required Notes
NVIDIA ❌ No Public API
OpenRouter ❌ No Public API
Cohere ✅ Yes Free tier available
DeepSeek ✅ Yes -
Fireworks ✅ Yes -
GitHub ✅ Yes Uses GITHUB_TOKEN
Groq ✅ Yes Free tier available
HuggingFace ✅ Yes Free tier available
Mistral ✅ Yes -
TogetherAI ✅ Yes -
xAI ✅ Yes -

Web Scraping Scripts

The following scripts use web scraping instead of official APIs and may break if the source website changes:

Provider Source URL Risk Level
NanoGPT nano-gpt.com/api Medium
Perplexity Documentation page High
SambaNova Community docs High

Recommendation: Monitor these scripts for failures and consider implementing official API calls when documentation becomes available.

Unverifiable Providers

The following providers could not be verified due to lack of public API documentation:

These scripts appear functional based on code review but cannot be validated against official specifications.

Contributing

When adding new providers:

  1. Create a new fetcher script in the scripts directory
  2. Update the provider list in update_models.py
  3. Add any required API keys to .env.example
  4. Document the API endpoint and any authentication requirements
  5. Prefer official APIs over web scraping when available