Cookie Preferences
We use cookies to personalize content and analyze traffic. See our Cookie Policy.
Include previous turns when sending the next request so your coding context is preserved.
Claude Code via ApiPass
Ask for code reviews, implementation plans, debugging help, or refactors using the Claude Code relay.
Explore different use cases and parameter configurations
Complete guide to using Claude Code
Run Claude Code-style coding conversations through ApiPass with the same account, billing, and API key workflow you already use for other models.
Claude Code on ApiPass is a browser-accessible coding assistant experience that routes requests through the same Claude Code relay infrastructure used elsewhere in the platform. Instead of requiring a local CLI session, this page lets you test prompts, evaluate replies, and validate API behavior directly from the model page while still using the OpenAI-compatible `/v1/chat/completions` contract with `model=claude-code`.
Use the page to ask Claude Code to inspect error symptoms, compare hypotheses, and outline likely causes before you wire it into your own tooling.
Validate prompts for architecture planning, migration steps, or boundary extraction before embedding the model into internal code assistants.
Check how Claude Code responds to review prompts, missing-test detection, and implementation summaries so your production integration has predictable output.
You can move from browser testing to application integration in a few predictable steps.
Claude Code requests use the same ApiPass API key system as other models, so you do not need a separate authentication flow for the browser page and your own application.
Try the same coding prompts you plan to send from your product. This helps confirm the level of detail, tone, and structure you want before integration.
Your backend or client can then send the same request shape shown in the API tab. This keeps browser experiments and production usage aligned.
Claude Code performs best when your prompt clearly describes the engineering task, the constraints, and the output format you want back.
Describe the bug, file, or behavior you want fixed instead of giving a broad 'improve this' request.
Leave conversation memory enabled if you want Claude Code to refine earlier answers or continue a debugging thread.
The same model can be called from your own client through ApiPass using the sample request in the API tab.
Teams can use Claude Code through ApiPass as the reasoning layer behind internal code review bots, debugging helpers, or architectural Q&A tools.
If your product needs implementation guidance, patch suggestions, or structured debugging responses, this model page gives you a fast way to validate prompt quality before shipping.
Claude Code can summarize stack traces, propose likely fault domains, and draft remediation steps, making it useful beyond pure code generation.
You can test Claude Code alongside other ApiPass models without building a separate auth and billing pipeline.
Prompt iteration happens directly on the model page, so you can tune your coding workflows before writing integration code.
The browser page and your own integration both target the same `claude-code` model contract, which reduces drift between testing and deployment.
All APIs require authentication via Bearer Token.
Authorization: Bearer
OpenAI-compatible chat completions endpoint backed by the Claude Code relay.
Send an OpenAI-compatible chat completions request.
1{
2 "model": "claude-code",
3 "stream": true,
4 "memory": true,
5 "messages": [
6 {
7 "role": "user",
8 "content": "Review this retry loop and suggest a safer implementation for transient API errors."
9 }
10 ]
11}1curl -X POST "https://api.apipass.dev/v1/chat/completions" \
2 -H "Content-Type: application/json" \
3 -H "Authorization: Bearer YOUR_API_KEY" \
4 -d '{
5 "model": "claude-code",
6 "stream": true,
7 "memory": true,
8 "messages": [
9 {
10 "role": "user",
11 "content": "Review this retry loop and suggest a safer implementation for transient API errors."
12 }
13 ]
14}'1{
2 "id": "chatcmpl_demo",
3 "object": "chat.completion",
4 "created": 1700000000,
5 "model": "claude-code",
6 "choices": [
7 {
8 "index": 0,
9 "finish_reason": "stop",
10 "message": {
11 "role": "assistant",
12 "content": "A safer retry loop should classify retryable errors first, then apply capped exponential backoff with jitter."
13 }
14 }
15 ],
16 "usage": {
17 "prompt_tokens": 186,
18 "completion_tokens": 73,
19 "total_tokens": 259
20 }
21}choices[0].message.contentAssistant reply content.
usage.prompt_tokensEstimated input token count.
usage.completion_tokensGenerated output token count.