9d5f4da3d6
Maple AI is a TEE-based private inference provider using Confidential Computing for end-to-end encryption with cryptographic attestations. - Add maple-models.ts with model catalog and discovery - Add MAPLE_API_KEY environment variable support - Add interactive onboarding: `clawdbot onboard --auth-choice maple-api-key` - Add configurable proxy URL (default: http://127.0.0.1:8080/v1) - Add provider auto-registration when API key detected - Add documentation at docs/providers/maple.md Supported models: kimi-k2-thinking (default), gpt-oss-120b, deepseek-r1-0528, qwen3-coder-480b, qwen3-vl-30b, llama-3.3-70b, gemma-3-27b
205 lines
6.1 KiB
Markdown
205 lines
6.1 KiB
Markdown
# Maple AI Provider
|
|
|
|
Maple AI provides TEE-based (Trusted Execution Environment) private inference using Confidential Computing. All inference runs in secure enclaves with end-to-end encryption and cryptographic attestations, ensuring your prompts and responses remain private.
|
|
|
|
## How It Works
|
|
|
|
Maple AI runs as a local proxy (desktop app or Docker container) that connects to secure TEE enclaves. Your data is encrypted end-to-end and never visible to Maple or any third party. Maple runs the largest, state-of-the-art open models and does not share any data back to the model creators. Sign up at [trymaple.ai](https://trymaple.ai) to get started. Maple Proxy requires a paid account with API credits.
|
|
|
|
1. **Desktop App or Docker**: Run the Maple proxy locally
|
|
2. **Local Proxy**: Default endpoint at `http://127.0.0.1:8080/v1`
|
|
3. **TEE Backend**: Requests are forwarded to Maple's secure enclaves
|
|
4. **Cryptographic Attestation**: Verify the enclave is running trusted code
|
|
|
|
## Features
|
|
|
|
- **End-to-end encryption**: Your prompts and responses are encrypted
|
|
- **Cryptographic attestations**: Verify the secure enclave integrity
|
|
- **Open-source verifiable code**: Audit the code running in the enclave
|
|
- **OpenAI-compatible API**: Standard `/v1` endpoints for easy integration
|
|
- **Streaming**: Required for all completions
|
|
|
|
## Setup
|
|
|
|
### 1. Install Maple Proxy
|
|
|
|
**Desktop App (Recommended)**
|
|
|
|
Download and run the Maple desktop app from [trymaple.ai/downloads](https://trymaple.ai/downloads). The proxy runs automatically on `http://127.0.0.1:8080/v1`.
|
|
|
|
**Docker**
|
|
|
|
```bash
|
|
docker run -p 8080:8080 \
|
|
-e MAPLE_BACKEND_URL=https://enclave.trymaple.ai \
|
|
-e MAPLE_ENABLE_CORS=true \
|
|
trymaple/proxy
|
|
```
|
|
|
|
### 2. Generate API Key
|
|
|
|
Open the Maple app and generate an API key. This key authenticates your requests to the local proxy.
|
|
|
|
### 3. Configure Moltbot
|
|
|
|
**Option A: Interactive Setup (Recommended)**
|
|
|
|
```bash
|
|
moltbot onboard --auth-choice maple-api-key
|
|
```
|
|
|
|
This will:
|
|
1. Prompt for your API key
|
|
2. Ask for the proxy URL (defaults to `http://127.0.0.1:8080/v1`)
|
|
3. Configure the provider automatically
|
|
|
|
**Option B: Environment Variable**
|
|
|
|
```bash
|
|
export MAPLE_API_KEY="your-api-key"
|
|
```
|
|
|
|
**Option C: Non-interactive**
|
|
|
|
```bash
|
|
moltbot onboard --non-interactive \
|
|
--auth-choice maple-api-key \
|
|
--token "your-api-key" \
|
|
--token-provider maple
|
|
```
|
|
|
|
### 4. Verify Setup
|
|
|
|
```bash
|
|
moltbot chat --model maple/llama-3.3-70b "Hello, are you working?"
|
|
```
|
|
|
|
## Available Models
|
|
|
|
| Model ID | Name | Use Case | Pricing |
|
|
|----------|------|----------|---------|
|
|
| `kimi-k2-thinking` | Kimi K2 Thinking | Complex agentic workflows, multi-step coding, web research | $4/$4 per M tokens |
|
|
| `gpt-oss-120b` | GPT OSS 120B | Creative writing, structured data | $4/$4 |
|
|
| `deepseek-r1-0528` | DeepSeek R1 | Research, advanced math, coding | $4/$4 |
|
|
| `qwen3-coder-480b` | Qwen3 Coder 480B | Agentic coding, large codebase analysis, browser automation | $4/$4 |
|
|
| `qwen3-vl-30b` | Qwen3 VL 30B | Image and video analysis, screenshot-to-code, OCR, GUI automation | $4/$4 |
|
|
| `llama-3.3-70b` | Llama 3.3 70B | General reasoning, conversation | $4/$4 |
|
|
| `gemma-3-27b` | Gemma 3 27B | General purpose, efficient | $10/$10 |
|
|
|
|
## Model Selection
|
|
|
|
Change your default model anytime:
|
|
|
|
```bash
|
|
moltbot models set maple/llama-3.3-70b
|
|
moltbot models set maple/deepseek-r1-0528
|
|
```
|
|
|
|
List available models:
|
|
|
|
```bash
|
|
moltbot models list | grep maple
|
|
```
|
|
|
|
## Configuration
|
|
|
|
### Custom Proxy URL
|
|
|
|
If running the proxy on a different host or port:
|
|
|
|
```yaml
|
|
# ~/.moltbot.yaml
|
|
models:
|
|
providers:
|
|
maple:
|
|
baseUrl: "http://192.168.1.100:8080/v1"
|
|
api: "openai-completions"
|
|
apiKey: "MAPLE_API_KEY"
|
|
```
|
|
|
|
### Docker Environment Variables
|
|
|
|
| Variable | Description | Default |
|
|
|----------|-------------|---------|
|
|
| `MAPLE_BACKEND_URL` | TEE enclave URL | `https://enclave.trymaple.ai` |
|
|
| `MAPLE_ENABLE_CORS` | Enable CORS headers | `false` |
|
|
| `RUST_LOG` | Log level | `info` |
|
|
|
|
## Usage Examples
|
|
|
|
```bash
|
|
# General chat
|
|
moltbot chat --model maple/llama-3.3-70b
|
|
|
|
# Advanced reasoning
|
|
moltbot chat --model maple/kimi-k2-thinking
|
|
|
|
# Research and coding
|
|
moltbot chat --model maple/deepseek-r1-0528
|
|
|
|
# Vision tasks
|
|
moltbot chat --model maple/qwen3-vl-30b
|
|
|
|
# Coding tasks
|
|
moltbot chat --model maple/qwen3-coder-480b
|
|
```
|
|
|
|
## Privacy and Security
|
|
|
|
### Why TEE?
|
|
|
|
Trusted Execution Environments (TEEs) provide hardware-level isolation:
|
|
|
|
- **Memory encryption**: Data is encrypted in memory
|
|
- **Attestation**: Cryptographic proof of what code is running
|
|
- **Isolation**: Even the host system cannot access enclave data
|
|
|
|
### Security Proof Attestation
|
|
|
|
Maple provides cryptographic attestations that prove the integrity of the secure enclave. You can view the current attestations at [trymaple.ai/proof](https://trymaple.ai/proof).
|
|
|
|
The Maple Proxy automatically verifies these attestations before connecting to the backend. If the attestation is invalid or tampered with, the proxy refuses to connect, similar to how SSL/TLS certificates protect web connections. This ensures you're always communicating with a genuine, unmodified Maple enclave.
|
|
|
|
### Verification
|
|
|
|
You can verify the enclave attestation to ensure:
|
|
|
|
1. The code running matches the open-source release
|
|
2. The TEE hardware is genuine
|
|
3. No tampering has occurred
|
|
|
|
Visit [trymaple.ai/proof](https://trymaple.ai/proof) to inspect the current attestation details.
|
|
|
|
## Troubleshooting
|
|
|
|
### Proxy not running
|
|
|
|
Ensure the Maple app is running or Docker container is active:
|
|
|
|
```bash
|
|
curl http://127.0.0.1:8080/health
|
|
```
|
|
|
|
### Connection refused
|
|
|
|
Check the proxy URL is correct and the service is running:
|
|
|
|
```bash
|
|
# Test connectivity
|
|
curl -H "Authorization: Bearer $MAPLE_API_KEY" \
|
|
http://127.0.0.1:8080/v1/models
|
|
```
|
|
|
|
### Model not available
|
|
|
|
The model list is fetched from the proxy. Ensure your Maple subscription includes the model you're trying to use.
|
|
|
|
## Links
|
|
|
|
- [Maple AI](https://trymaple.ai)
|
|
- [Downloads](https://trymaple.ai/downloads)
|
|
- [Security Proof](https://trymaple.ai/proof)
|
|
- [Proxy Documentation](https://blog.trymaple.ai/maple-proxy-documentation/)
|
|
- [Maple GitHub](https://github.com/OpenSecretCloud/Maple)
|
|
- [Maple Proxy GitHub](https://github.com/OpenSecretCloud/maple-proxy)
|