Description

Media understanding providers (audio transcription with Groq/OpenAI, image/video description) don't use any configured proxy settings. This causes API calls to fail in environments that require a proxy for internet access.

Steps to Reproduce

1. Configure environment variable proxy (not used by media providers):

   export HTTPS_PROXY=http://127.0.0.1:10808
   export HTTP_PROXY=http://127.0.0.1:10808

2. Or configure in openclaw.json (also not used):

{
  "env": {
    "HTTPS_PROXY": "http://127.0.0.1:10808",
    "HTTP_PROXY": "http://127.0.0.1:10808"
  }
}

2. Configure Telegram proxy (works for Telegram, but not for media providers):

{
  "channels": {
    "telegram": {
      "proxy": "http://127.0.0.1:10808"
    }
  }
}

3. Configure audio transcription:

{
  "tools": {
    "media": {
      "audio": {
        "enabled": true,
        "models": [{
          "provider": "groq",
          "model": "whisper-large-v3-turbo"
        }]
      }
    }
  },
  "env": {
    "GROQ_API_KEY": "gsk-..."
  }
}

4. Send a voice message to Telegram bot

Expected Behavior

Audio transcription should work. The Groq API call (https://api.groq.com/openai/v1/audio/transcriptions) should use one of the configured proxy settings:

• Environment variables HTTPS_PROXY / HTTP_PROXY
• Or a dedicated media proxy configuration

Actual Behavior

• Telegram getFile API works (uses grammY's configured proxy from channels.telegram.proxy)
• Groq API call fails (doesn't use environment variables proxy, doesn't use Telegram proxy)
• Agent receives media:audio placeholder instead of transcribed text
• No error messages (silent failure)

Only workaround: Enable system-level TUN mode proxy to intercept all network traffic at OS level.

Root Cause

In src/media-understanding/runner.ts:925-937, when calling provider.transcribeAudio():

const result = await provider.transcribeAudio({
  buffer: media.buffer,
  fileName: media.fileName,
  mime: media.mime,
  apiKey,
  baseUrl,
  headers,
  model,
  language: ...,
  prompt,
  query: providerQuery,
  timeoutMs,
  // ❌ No fetchFn parameter passed
});

Then in src/media-understanding/providers/openai/audio.ts:16:

const fetchFn = params.fetchFn ?? fetch;
// Uses global fetch, which doesn't respect HTTPS_PROXY/HTTP_PROXY env vars

Issue:

• Node.js's undici fetch (used by OpenClaw) doesn't automatically read HTTPS_PROXY / HTTP_PROXY environment variables
• Media understanding providers have no way to configure proxy
• Telegram proxy is only applied to grammY client, not to media provider API calls