windows ai project recovery agent

The problem

Local Windows AI projects often fail due to port conflicts, broken Python virtual environments, or mismatched Ollama configurations. Developers waste hours debugging "Unexpected token <" errors and cryptic BAT script failures that close windows before logs can be read.

What it does

• Identifies blocked ports and the specific PIDs holding them using PowerShell diagnostics.
• Validates Python environments and ensures the correct interpreter and dependencies are active.
• Tests Ollama API health and verifies model availability, tags, and connection timeouts.
• Inspects raw API responses to fix common "HTML-instead-of-JSON" frontend failures.
• Audits BAT and PowerShell launchers for path errors, drive mismatches, and improper command chaining.

Frameworks & tools

Windows 10/11, Python (Flask, FastAPI), Node.js, Ollama, PowerShell, and BAT scripts.

Why this beats prompting it yourself

A standard LLM often suggests destructive fixes like reinstalling everything. This skill follows a strict "diagnose before modify" protocol, creating reversible backups and verifying network bindings before touching a single line of code.

Use cases

• Recovering a local AI dashboard that fails to connect to the backend after a Windows update.
• Fixing "model not found" or 404 errors in Ollama-based applications.
• Debugging Python startup failures where the terminal window closes too fast to see errors.
• Resolving CORS or port binding issues in Flask/FastAPI and Node.js stacks.

Known limitations

Designed specifically for Windows environments. It does not perform destructive actions like firewall rule changes or package installs without explicit user confirmation.