在opencode虽然能调用，但是会报错

[AIGeekMaker]

opencode需要这么配置cc
baseURL": "http://localhost:23000/v1

是cch的问题还是中转站的问题？

[coderabbitai]

📝 CodeRabbit Plan Mode

Generate an implementation plan and prompts that you can use with your favorite coding agent.

• Create Plan

Examples

• Example 1
• Example 2

---

⚠️ Possible Duplicate Issue

• [bug] 用 codex (Response API) 時, 如果base_url本身不需要有v1的, 都會被加上v1導致請求出錯 #139

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[github-actions]

Thank you for reporting this issue.

Analysis:

Based on my investigation, this appears to be related to a previously reported issue #139 which described the same problem - the proxy was adding /v1 to base URLs that already contained the full path. That issue was closed with a note that it would be fixed in a later version.

Good news: The fix for this issue appears to already be implemented in the current codebase.

Relevant code:

The buildProxyUrl() function in src/app/v1/_lib/url.ts (lines 25-84) now implements smart path detection:

// Smart detection: Check if base_url already contains the complete target path
const targetEndpoints = [
  "/responses",        // Codex Response API
  "/messages",         // Claude Messages API
  "/chat/completions", // OpenAI Compatible
  "/models",           // Gemini & OpenAI models
];

The function checks whether the configured base_url already ends with these endpoints (lines 41-65), and if so, skips path concatenation to avoid generating URLs like https://xxx.com/openai/v1/responses/v1/responses.

To diagnose your specific error:

1. Check your provider configuration - In CCH dashboard, verify:

   • What is the exact base_url you configured for the provider?
   • What provider_type did you select? (should be openai-compatible or codex for OpenCode)

2. Please share the error message - The screenshots in your issue are not viewable as text. Could you provide:

   • The exact error message text from OpenCode
   • The request URL that CCH is generating (visible in CCH logs)

3. Configuration for OpenCode - Your baseURL format http://localhost:23000/v1 looks correct if CCH is running on port 23000. The endpoint should be:

   • /v1/chat/completions for OpenAI-compatible format
   • /v1/responses for Response API format

Potential workaround:

If you're still experiencing issues, try configuring your provider's base_url in CCH without the /v1 prefix, as CCH should automatically append it based on the request path.

---

Automated response from Claude AI

[hank9999]

中转站的 CC 渠道一般仅限在 Claude Code 中使用, 建议绕过 CCH 配置试一下

[ding113]

您好，这是因为 opencode 中模型名称错误。请检查 opencode 配置，确保使用带日期后缀的模型名称，如
claude-sonnet-4-5-20250929

[LinAsan-1205]

"models": {
"claude-sonnet-4-5-20250929": {
"name": "Sonnet",
"limit": {
"context": 200000,
"output": 65536
}
},
"claude-haiku-4-5-20251001": {
"name": "Haiku",
"limit": {
"context": 200000,
"output": 65536
}
},
"claude-opus-4-5-20251101": {
"name": "Opus",
"limit": {
"context": 200000,
"output": 65536
}
}
} 这样可以用

[rexleimo]

openai 那边有问题