Skip to content
Back to skills

2556 Api Usage 0fa5b854

ASecurity

The CCProxy API is a reverse proxy to api.anthropic.com that provides both Anthropic and OpenAI-compatible interfaces. It offers two main access modes:

  • 9 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 11, 2026
ai-agentsbashgitapi

Works with

  • claude code
  • api
  • mcp

Security analysis

A100/100

Scanned October 11, 2026

npx -y skills add tools-only/X-Skills --skill 2556-api-usage_0fa5b854 --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of 2556 Api Usage 0fa5b854?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for 2556 Api Usage 0fa5b854
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/tools-only-2556-api-usage-0fa5b854/badge)](https://www.skillsdirectory.com/skills/tools-only-2556-api-usage-0fa5b854)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

SKILL.md
# API Usage

## Overview

The CCProxy API is a reverse proxy to api.anthropic.com that provides both Anthropic and OpenAI-compatible interfaces. It offers two main access modes:

## Access Modes

| Mode | URL Prefix | Method | Use Case |
|------|------------|--------|----------|
| Claude Code | `/` or `/cc/` | Uses claude-code-sdk | Access to all Claude Code tools |
| API | `/api/` | Direct proxy with headers | Full API control and settings |

## Anthropic API Format

### Base URLs by Mode
```
Claude SDK Mode: http://localhost:8000/claude/sdk/v1/
Claude API Mode: http://localhost:8000/claude/v1/
```

### Messages Endpoint
```bash
# Claude SDK mode - with all tools
curl -X POST http://localhost:8000/claude/sdk/v1/messages \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-3-5-sonnet-20241022",
    "max_tokens": 1000,
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'

# Claude API mode - direct proxy for full control
curl -X POST http://localhost:8000/claude/v1/messages \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-3-5-sonnet-20241022",
    "max_tokens": 1000,
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'
```

## OpenAI API Format

### Base URLs by Mode
```
Claude SDK Mode: http://localhost:8000/claude/sdk/v1/
Claude API Mode: http://localhost:8000/claude/v1/
Codex (OpenAI):  http://localhost:8000/codex/v1/
Copilot (GitHub): http://localhost:8000/copilot/v1/
```

### Chat Completions
```bash
# Claude SDK mode - with all tools
curl -X POST http://localhost:8000/claude/sdk/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-3-5-sonnet-20241022",
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'

# Claude API mode - direct proxy for full control
curl -X POST http://localhost:8000/claude/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-3-5-sonnet-20241022",
    "messages": [
      {"role": "user", "content": "Hello, Claude!"}
    ]
  }'

### Codex (OpenAI) Examples

```bash
# OpenAI Chat Completions via Codex
curl -X POST http://localhost:8000/codex/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5",
    "messages": [
      {"role": "user", "content": "Hello from Codex!"}
    ]
  }'

# OpenAI Responses API via Codex
curl -X POST http://localhost:8000/codex/v1/responses \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5",
    "input": [
      {"role": "user", "content": [{"type": "text", "text": "Hello!"}]}
    ]
  }'

# OpenAI Chat Completions via Copilot
curl -X POST http://localhost:8000/copilot/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4-copilot",
    "messages": [
      {"role": "user", "content": "Hello from Copilot!"}
    ]
  }'

# OpenAI Responses API via Copilot
curl -X POST http://localhost:8000/copilot/v1/responses \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4-copilot",
    "input": [
      {"role": "user", "content": [{"type": "text", "text": "Hello!"}]}
    ]
  }'
```
```

## Supported Models

- claude-3-5-sonnet-20241022
- claude-3-5-haiku-20241022
- claude-3-opus-20240229
- claude-3-sonnet-20240229
- claude-3-haiku-20240307

## Function Calling and Tools

### Claude Code Mode Limitations
When using Claude Code mode (default), the proxy uses the claude-code-sdk which has these limitations:
- Cannot directly use ToolCall/function calling through the API
- Limited control over model settings
- However, you get access to all tools configured in Claude Code

### API Mode Advantages
When using API mode (`/api/*`), you get:
- Full access to all API features including ToolCall
- Complete control over model settings
- Direct pass-through to api.anthropic.com

### MCP Integration
You can extend Claude's capabilities using MCP (Model Context Protocol) servers in Claude Code mode.

For detailed information on setting up and using MCP servers with Claude Code, see the [MCP Server Integration guide](mcp-integration.md).

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.