LM Studio Powershell Proxy

Route any OpenAI-compatible client through a local PowerShell proxy to your LM Studio server. This README covers setup, hosts changes, and execution (no source code here).

---

1) Prerequisites

• Windows 10/11
• LM Studio running its OpenAI-compatible server (default: http://127.0.0.1:1234/v1)

# Verify LM Studio is up:
Invoke-RestMethod http://127.0.0.1:1234/v1/models

---

2) Place the proxy script

Put your lmproxy.ps1 (the proxy script) somewhere convenient, e.g.:

C:\Users\<YOUR_USER>\Desktop\lmproxy.ps1

The script should forward POST /v1/chat/completions to http://127.0.0.1:1234/v1/chat/completions and return a fake GET /v1/models list that includes your model ID (e.g. qwen/qwen3-coder-30b).

---

3) One-time URL ACL for port 8080

Run PowerShell as Administrator and grant the URL reservation:

netsh http add urlacl url=http://+:8080/ user=Everyone

---

4) Hosts file change (point OpenAI → your proxy)

From Administrator PowerShell:

Add-Content -Path "$env:SystemRoot\System32\drivers\etc\hosts" -Value "`n127.0.0.1 api.openai.com"
ipconfig /flushdns

Check:

Select-String -Path "$env:SystemRoot\System32\drivers\etc\hosts" -Pattern "api.openai.com"

To undo later:

(Get-Content "$env:SystemRoot\System32\drivers\etc\hosts") |
  Where-Object {$_ -notmatch '^127\.0\.0\.1\s+api\.openai\.com$'} |
  Set-Content "$env:SystemRoot\System32\drivers\etc\hosts"
ipconfig /flushdns

---

5) Run the proxy

Start as Administrator so it can bind to port 8080:

powershell -ExecutionPolicy Bypass -File "C:\Users\<YOUR_USER>\Desktop\lmproxy.ps1"

You should see something like:

Proxy listening on http://+:8080  ->  http://127.0.0.1:1234

---

6) Quick tests (through the proxy)

Models (fake list from the proxy):

Invoke-RestMethod -Uri "http://api.openai.com:8080/v1/models" -Headers @{Authorization="Bearer sk-local-test"}

Chat completion passthrough:

$body = @{
  model    = "qwen/qwen3-coder-30b"   # must match what the proxy returns at /v1/models
  messages = @(@{ role = "user"; content = "Say hello from LM Studio." })
} | ConvertTo-Json -Depth 5

Invoke-RestMethod -Uri "http://api.openai.com:8080/v1/chat/completions" `
  -Method Post `
  -Headers @{Authorization="Bearer sk-local-test"; "Content-Type"="application/json"} `
  -Body $body

---

7) Point your editor/tool at the proxy

Use these settings in any OpenAI-compatible client:

• Base URL: http://api.openai.com:8080/v1
• API Key: any placeholder (e.g. sk-local-test)
• Model: the exact ID your proxy returns at /v1/models (e.g. qwen/qwen3-coder-30b)

Note: In Cursor free tiers, only Ask/Chat will work. Agent/Edit are paywalled by Cursor.

---

8) Troubleshooting

• Access denied on start: make sure you ran the netsh http add urlacl ... command and are starting PowerShell as Administrator.
• No response / 404: your proxy likely only handles /v1/models and /v1/chat/completions. Check the endpoint path.
• Model not found: ensure the model ID in your client exactly matches what your proxy returns at /v1/models.
• LM Studio not responding: confirm http://127.0.0.1:1234/v1/models works and that your model is loaded.

---