Error 500 (no body) on all requests

[hugomart-inria]

Hello,
I had a config.yaml working just fine with Chat Plan and Agent, but recently (a week ?), I have had this error on each prompt. I use a Qwen3-30B model. In the logs, there are messages about config.yaml not getting seen by the AST tracker. I tried going back to older version of Continue but it changed nothing. My API is accessible through other services.

[xXMrNidaXx]

500 errors with no body are frustrating to debug! Here's a systematic approach:

1. Check server logs

# Find Continue's server process
ps aux | grep continue

# Check its output
tail -f ~/.continue/logs/server.log

2. Enable verbose logging

// config.json
{
  "logging": {
    "level": "debug"
  }
}

3. Test the backend directly

curl -X POST http://localhost:65432/health
curl -X POST http://localhost:65432/v1/completions \
  -H "Content-Type: application/json" \
  -d '{"prompt": "test"}' -v

4. Common causes:

• Model provider is down/unreachable
• API key invalid or expired
• Request too large (context overflow)
• SSL/TLS issues with provider
• Firewall blocking outbound

5. Isolate the issue

# Test provider directly
curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer $OPENAI_API_KEY"

6. Reset state

# Clear Continue cache
rm -rf ~/.continue/cache
# Restart extension

We've debugged these at RevolutionAI. Usually it's an auth or network issue.

What provider are you using, and what triggers the 500?