Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/finesssee/proxypilot

Windows-native CLIProxyAPI fork with TUI, system tray, and multi-provider OAuth for AI coding tools - No API key needed
https://github.com/finesssee/proxypilot

claude-code cli-proxy codex factory-droids gemini kiro

Last synced: about 1 month ago
JSON representation

Windows-native CLIProxyAPI fork with TUI, system tray, and multi-provider OAuth for AI coding tools - No API key needed

Awesome Lists containing this project

README 

          


  ProxyPilot Logo




ProxyPilot





  MIT License

  Go Version

  Follow on 𝕏

  Star this repo






  Stop juggling API keys. ProxyPilot is a powerful local API proxy that lets you use your existing Claude Code, Codex, Gemini, Kiro, and Qwen subscriptions with any AI coding tool – no separate API keys required.






  Built in Go, it handles OAuth authentication, token management, and API translation automatically. One server to route them all.




---



> [!TIP]

> πŸ“£ **Latest models supported:**

> Claude Opus 4.5 / Sonnet 4.5 with extended thinking, GPT-5.2 / GPT-5.2 Codex, Gemini 3 Pro/Flash, and Kiro (AWS CodeWhisperer)! πŸš€



**Setup Guides:**

- [Claude Code Setup β†’](docs/claude-code-local-proxy.md)

- [Cursor IDE Setup β†’](docs/cursor-ide.md)



---



## Features



- 🎯 **10 Auth Providers** - Claude, Codex (OpenAI), Gemini, Gemini CLI, Kiro (AWS), Amazon Q CLI, Qwen, Antigravity, MiniMax, Zhipu AI

- πŸ”„ **Universal API Translation** - Auto-converts between OpenAI, Anthropic, and Gemini formats

- πŸ”§ **Tool Calling Repair** - Fixes tool/function call mismatches between providers automatically

- 🧠 **Extended Thinking** - Full support for Claude and Gemini thinking models

- πŸ” **OAuth Integration** - Browser-based login with automatic token refresh

- πŸ‘₯ **Multi-Account Support** - Round-robin distribution with automatic failover

- ⚑ **Quota Auto-Switch** - Automatically switches to backup project/model when quota exceeded

- πŸ“Š **Usage Statistics** - Track requests, tokens, and errors per provider/model

- 🧩 **Context Compression** - LLM-based summarization for long sessions (Factory.ai research)

- πŸ€– **Agentic Harness** - Guided workflow for coding agents (Anthropic research)

- πŸ’Ύ **Session Memory** - Persistent storage across conversation turns

- 🎨 **System Tray** - Native Windows tray app for quick access

- πŸ”„ **Auto-Updates** - Background update checking with one-click install

- βͺ **Rollback Support** - Automatic crash detection with version recovery

- πŸ“‘ **60+ Management APIs** - Full control via REST endpoints



---



## Supported Providers



| Provider | Auth Method | Models |

|----------|-------------|--------|

| Claude (Anthropic) | OAuth2 / API Key | Claude Opus 4.5, Sonnet 4.5, Haiku 4.5 |

| Codex (OpenAI) | OAuth2 / API Key | GPT-5.2, GPT-5.2 Codex |

| Gemini | OAuth2 / API Key | Gemini 3 Pro, Gemini 3 Flash |

| Gemini CLI | OAuth2 | Cloud Code Assist (separate quota) |

| Kiro | OAuth2 + AWS SSO | AWS CodeWhisperer |

| Amazon Q CLI | Import from CLI | Amazon Q Developer |

| Qwen | OAuth2 | Qwen models |

| Antigravity | OAuth2 | Gemini via Antigravity (separate quota) |

| MiniMax | API Key | MiniMax M2, M2.1 models |

| Zhipu AI | API Key | GLM-4.5, GLM-4.6, GLM-4.7 |

| Custom | API Key | Any OpenAI-compatible endpoint |



---



## Installation



### Download Pre-built Release (Recommended)



1. Go to the [**Releases**](https://github.com/Finesssee/ProxyPilot/releases) page

2. Download the latest binary for your platform

3. Run `./proxypilot`



### Build from Source



```bash

git clone https://github.com/Finesssee/ProxyPilot.git

cd ProxyPilot

go build -o proxypilot ./cmd/server

./proxypilot

```



---



## Usage



### First Launch



1. Copy config: `cp config.example.yaml config.yaml`

2. Run: `./proxypilot`

3. Server starts on `http://localhost:8317`

4. Open dashboard: `http://localhost:8317/proxypilot.html`



### Authentication



Run OAuth login for your provider:



```bash

# OAuth providers (opens browser)

./proxypilot --claude-login        # Claude

./proxypilot --codex-login         # OpenAI/Codex

./proxypilot --login               # Gemini

./proxypilot --kiro-login          # Kiro (Google OAuth)

./proxypilot --kiro-aws-login      # Kiro (AWS Builder ID)

./proxypilot --qwen-login          # Qwen

./proxypilot --antigravity-login   # Antigravity



# Import providers (from existing CLI tools)

./proxypilot --kiro-import         # Kiro IDE token



# API key providers (prompts for key)

./proxypilot --minimax-login       # MiniMax API key

./proxypilot --zhipu-login         # Zhipu AI API key

```



OAuth tokens are stored locally and auto-refreshed before expiry.



### Security Defaults (Auth + CORS)



- Proxy requests require API keys by default. To allow unauthenticated access (not recommended), set `allow-unauthenticated: true` in `config.yaml`.

- CORS is enabled for non-management endpoints by default (wildcard `*`). Management endpoints **do not** emit CORS headers unless you explicitly allow origins under `cors.management-allow-origins`.



Example:



```yaml

allow-unauthenticated: false

cors:

  allow-origins:

    - "http://localhost:5173"

  management-allow-origins:

    - "http://localhost:5173"

```



### Configure Your Tools



**Claude Code** (`~/.claude/settings.json`):

```json

{

  "env": {

    "ANTHROPIC_BASE_URL": "http://127.0.0.1:8317",

    "ANTHROPIC_AUTH_TOKEN": "your-api-key"

  }

}

```



**Codex CLI** (`~/.codex/config.toml`):

```toml

[openai]

api_base_url = "http://127.0.0.1:8317"

```



**Factory Droid** (`~/.factory/settings.json`):

```json

{

  "customModels": [{

    "name": "ProxyPilot",

    "baseUrl": "http://127.0.0.1:8317"

  }]

}

```



---



## API Endpoints



```

POST /v1/chat/completions     # OpenAI Chat Completions

POST /v1/responses            # OpenAI Responses API

POST /v1/messages             # Anthropic Messages API

GET  /v1/models               # List available models

GET  /healthz                 # Health check

```



All endpoints auto-translate between formats based on the target provider.



---



## Caching



ProxyPilot includes two caching layers to reduce latency and token usage.



### Response Cache



Caches full API responses for identical requests. Useful for repeated queries during development.



**Config** (`config.yaml`):

```yaml

response-cache:

  enabled: true           # Default: false

  max-size: 1000          # Max entries (default: 1000)

  max-bytes: 0            # Optional total cache size cap in bytes

  ttl-seconds: 300        # Cache TTL (default: 300 = 5 min)

  exclude-models:         # Models to skip (supports wildcards)

    - "*-thinking"

    - "o1-*"

```



### Prompt Cache



Synthetic prompt caching for providers without native support. Tracks repeated system prompts and estimates token savings.



**Config** (`config.yaml`):

```yaml

prompt-cache:

  enabled: true           # Default: false

  max-size: 500           # Max entries (default: 500)

  max-bytes: 0            # Optional total cache size cap in bytes

  ttl-seconds: 1800       # Cache TTL (default: 1800 = 30 min)

```



### Cache Management API



| Endpoint | Method | Description |

|----------|--------|-------------|

| `/v0/management/cache/stats` | GET | Response cache stats (hits, misses, size) |

| `/v0/management/cache/clear` | POST | Clear response cache |

| `/v0/management/cache/enabled` | PUT | Enable/disable at runtime `{"enabled": true}` |

| `/v0/management/prompt-cache/stats` | GET | Prompt cache stats + estimated tokens saved |

| `/v0/management/prompt-cache/clear` | POST | Clear prompt cache |

| `/v0/management/prompt-cache/enabled` | PUT | Enable/disable at runtime |     

| `/v0/management/prompt-cache/top` | GET | Top 10 most-hit prompts |



---



## Lightweight Profile



For low‑memory or high‑throughput setups, you can disable heavier features:



```yaml

commercial-mode: true

usage-statistics-enabled: false

usage-sample-rate: 0.25

metrics-enabled: false

request-history-enabled: false

request-history-sample-rate: 0.25

agentic-harness-enabled: false

prompt-budget-enabled: false

request-log: false

response-cache:

  enabled: false

prompt-cache:

  enabled: false

```



---



## Auto-Updates



ProxyPilot includes a built-in auto-update system that checks for new releases and allows one-click installation.



### Configuration



**Config** (`config.yaml`):

```yaml

updates:

  auto-check: true              # Enable background update checking (default: true)

  check-interval-hours: 24      # How often to check (default: 24)

  notify-on-update: true        # Show toast notification when update available (default: true)

  channel: stable               # Update channel: "stable" or "prerelease" (default: stable)

```



### Features



- **Background Polling** - Checks for updates at configurable intervals

- **Dashboard Banner** - Proactive notification when update is available

- **One-Click Install** - Download, verify, and install from the tray menu

- **Session Dismissal** - Dismissed banners won't reappear until next session



### Tray Menu



When an update is available:

- **Settings β†’ Download & Install vX.X.X** - One-click update flow

- **Settings β†’ Check for Updates** - Manual check



---



## Rollback Support



ProxyPilot automatically backs up the previous version during updates and can restore it if something goes wrong.



### Automatic Recovery



- **Crash Detection** - Tracks rapid restarts within a 30-second window

- **Auto-Rollback** - After 3 rapid failures, automatically restores previous version

- **Health Marking** - After 30 seconds of stable operation, clears the failure counter



### Manual Rollback



From the tray menu:

- **Settings β†’ Rollback to Previous Version** - Restore the previous version



### How It Works



1. During update, current binary is saved as `.old` backup

2. Rollback metadata stored in `%APPDATA%/ProxyPilot/updates/`

3. On crash loop detection, previous version is automatically restored

4. After successful startup (30s), backup can be cleaned up



---



## CLI Tools



| Binary | Description |

|--------|-------------|

| `proxypilot` | Main CLI with server and config switching |

| `proxypilot-tray` | System tray app |



---



## Tool Integrations



Works with these AI coding tools:



- **Claude Code** - Auto-configure via settings.json

- **Codex CLI** - Auto-configure via config.toml

- **Factory Droid** - Auto-configure via settings.json

- **Cursor IDE** - Manual endpoint configuration

- **Continue** - Manual endpoint configuration



---



## Requirements



- macOS, Linux, or Windows

- Go 1.24+ (for building from source)



---



## Author



ProxyPilot is developed and maintained by [@Finesssee](https://github.com/Finesssee). Some contributors shown in the git history are from upstream [CLIProxyAPI](https://github.com/router-for-me/CLIProxyAPI) merges. Direct contributors to ProxyPilot will be listed here as the project grows.



---



## Credits



ProxyPilot builds upon excellent work from the open-source community:



- **[CLIProxyAPI](https://github.com/router-for-me/CLIProxyAPI)** - The original unified proxy server that inspired this project

- **[VibeProxy](https://github.com/automazeio/vibeproxy)** - Native macOS menu bar app showcasing clean proxy UX



Long-context features are inspired by research from the AI community:



- **[Factory.ai](https://factory.ai/news/evaluating-compression)** - Context compression techniques for long-running coding agents

- **[Anthropic](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents)** - Effective harnesses for long-running agents



Special thanks to these teams for sharing their work and insights.



---



## License



MIT License - see [LICENSE](LICENSE) for details.



---



## Support



- **Report Issues**: [GitHub Issues](https://github.com/Finesssee/ProxyPilot/issues)



---



## Disclaimer



This project is for educational and interoperability research purposes. It interacts with various APIs to provide compatibility layers.



- **Use at your own risk.** Authors are not responsible for account suspensions or service interruptions.

- **Not affiliated** with Google, OpenAI, Anthropic, Amazon, or any other provider.

- Users must comply with the Terms of Service of connected platforms.