[BUG] Cannot fetch openrouter models in WSL when https_proxy is set (AxiosError ERR_BAD_REQUEST)

[npnh2]

Problem (one or two sentences)

When using Roo-Code extension in a WSL environment with the https_proxy environment variable set, the plugin fails to fetch OpenRouter models.

Context (who is affected and when)

Reproduction steps

1. Environment.
   vscode 1.107.1 on Windows11
   Roo-Code extension v3.37.1 on WSL 2.6.1.0
   WSL environment variables:
   https_proxy=http://172.28.16.1:10810
   http_proxy=http://172.28.16.1:10810
2. Exact actions.
   Open a new vscode window and connect to wsl, then open Roo Code extension.
   Select API Provider OpenRouter, click Model list and there is nothing.
   If there is a cached model list, then rm -r $HOME/.vscode-server/data/User/globalStorage/rooveterinaryinc.roo-cline/ in wsl and retry.

Expected result

Expected getting a list of openrouter models.

Actual result

AxiosError ERR_BAD_REQUEST

Variations tried (optional)

No response

App Version

3.37.1

API Provider (optional)

OpenRouter

Model Used (optional)

No response

Roo Code Task Links (optional)

No response

Relevant logs or errors (optional)

ERR [Extension Host] Error fetching OpenRouter models: {
  "stack": "AxiosError: Request failed with status code 400\n\tat bB (/home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:172:1223)\n\tat IncomingMessage.<anonymous> (/home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:177:11196)\n\tat IncomingMessage.emit (node:events:531:35)\n\tat endReadableNT (node:internal/streams/readable:1698:12)\n\tat process.processTicksAndRejections (node:internal/process/task_queues:90:21)\n\tat iie.request (/home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:179:2262)\n\tat process.processTicksAndRejections (node:internal/process/task_queues:105:5)\n\tat async mEi (/home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:2808:9229)\n\tat async LGi (/home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:3186:14371)\n\tat async /home/xxx/.vscode-server/extensions/rooveterinaryinc.roo-cline-3.37.1/dist/extension.js:3186:16817",
  "message": "Request failed with status code 400",
  "name": "AxiosError",
  "code": "ERR_BAD_REQUEST",
  "config": {},
  "status": 400
}

I believe this issue has the same root cause as #872, which was closed because Cant repro:

The getOpenRouterModels function in https://github.com/RooCodeInc/Roo-Code/blob/main/src/api/providers/fetchers/openrouter.ts uses axios to make request and does not automatically support HTTP CONNECT tunneling when it detects https_proxy.

[roomote]

I've analyzed this issue and here's my proposed implementation plan:

Problem Analysis

The issue occurs because axios doesn't automatically support HTTP CONNECT tunneling when https_proxy/http_proxy environment variables are set. When making HTTPS requests through an HTTP proxy, axios needs to be configured with an httpsAgent that handles the CONNECT tunneling.

Proposed Solution

1. Add the https-proxy-agent package to the project dependencies. This is a well-maintained package that creates an HTTP agent capable of tunneling HTTPS requests through an HTTP proxy.

2. Create a shared axios proxy utility at src/utils/axiosProxy.ts that:

   • Detects https_proxy, HTTPS_PROXY, http_proxy, or HTTP_PROXY environment variables
   • Creates an HttpsProxyAgent when a proxy is configured
   • Exports either a preconfigured axios instance or a helper function to get axios config with proxy support

3. Update all axios-based model fetchers to use the proxy-aware axios configuration:

   • src/api/providers/fetchers/openrouter.ts (the main affected file)
   • src/api/providers/fetchers/deepinfra.ts
   • src/api/providers/fetchers/litellm.ts
   • src/api/providers/fetchers/io-intelligence.ts
   • src/api/providers/fetchers/vercel-ai-gateway.ts
   • src/api/providers/fetchers/chutes.ts
   • src/api/providers/fetchers/requesty.ts
   • src/api/providers/fetchers/huggingface.ts
   • src/api/providers/fetchers/unbound.ts
   • src/api/providers/fetchers/ollama.ts
   • src/api/providers/fetchers/lmstudio.ts

4. Add tests for the proxy utility to ensure it correctly handles:

   • When proxy environment variables are set
   • When proxy environment variables are not set
   • Both uppercase and lowercase environment variable names

Benefits

• Respects standard proxy environment variables used in Linux/WSL environments
• Minimal changes to existing code - just import and use the proxy-aware axios config
• Consistent proxy handling across all model fetchers
• No changes to user-facing configuration required

Please mention @roomote in a comment saying that you approve this plan, and I'll proceed with the implementation.