Skip to content

Troubleshooting

Jump to bottom
Traliran edited this page Aug 12, 2026 · 1 revision

❓ Troubleshooting & FAQ

This page addresses common connection errors, CORS blockages, database sync issues, and general frequently asked questions when using Traliran AI Hub.

🚨 Common Error Messages & Solutions

1. "Failed to Fetch" or Network Error

  • Cause: Usually caused by Cross-Origin Resource Sharing (CORS) restriction or Mixed Content blocking in your browser.
  • Solution for Cloud APIs: Verify that your API key is correct and your billing account has active credits (especially for OpenAI, DeepSeek, or Anthropic).
  • Solution for Local Models: Ensure your local server (Ollama or llama.cpp) is running and configured to allow cross-origin requests.

2. Mixed Content Error (HTTP vs HTTPS)

  • Cause: If you open Traliran AI Hub via HTTPS (e.g., official GitHub Pages deployment) and try to connect to a local HTTP endpoint (e.g., http://localhost:11434), modern browsers block the connection for security.
  • Solution:
    1. Run Traliran AI Hub locally using a simple HTTP server (like python -m http.server 8080) and access it via http://localhost:8080.
    2. Or set up a local reverse proxy (like Caddy or Nginx) with local TLS certificates for your local LLM service.

3. Localhost CORS Issues (Ollama / llama.cpp)

  • Ollama Fix: You must pass the OLLAMA_ORIGINS environment variable before starting Ollama. Run export OLLAMA_ORIGINS="*" before launching ollama serve.
  • llama.cpp Fix: Start llama-server with the --cors "*" flag enabled.

4. PocketBase Sync Fails / Unauthorized

  • Cause: Incorrect API rules or missing CORS setup in PocketBase settings.
  • Solution:
    1. Open PocketBase Admin Dashboard -> Settings -> CORS Settings, and add your app's URL.
    2. Check the API Rules on your chats, prompts, and settings collections to ensure list, view, create, and update actions are allowed for your authentication state.

🙋 Frequently Asked Questions (FAQ)

Where are my API keys stored?

Your API keys are stored strictly inside your browser's LocalStorage. They are never transmitted to any third-party backend or middleman server. All requests are sent directly from your browser to the chosen provider.

Is my chat history private?

Yes. Everything is processed and kept on your local client side. If you connect PocketBase, history syncs exclusively to your private, self-hosted database instance.

Why is there no build step or node_modules?

Traliran AI Hub is designed around the KISS principle (Keep It Simple, Stupid). Using modern ES6 modules and native Web APIs eliminates project bloat, dependency vulnerabilities, and complex compilation steps.

Can I use custom or fine-tuned GGUF models?

Yes. Load your model into Ollama or llama.cpp as usual, and select it from the model dropdown in Traliran AI Hub once connected.

Clone this wiki locally