@ebowwa/mcp-modal 0.1.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +96 -0
- package/dist/index.js +92728 -0
- package/mcp.json +10 -0
- package/package.json +62 -0
- package/tsconfig.json +31 -0
package/README.md
ADDED
|
@@ -0,0 +1,96 @@
|
|
|
1
|
+
# @ebowwa/mcp-modal
|
|
2
|
+
|
|
3
|
+
Modal GPU Cloud MCP server - manage GPU workloads, training, and robot inference.
|
|
4
|
+
|
|
5
|
+
## Features
|
|
6
|
+
|
|
7
|
+
- **App Management**: Create and manage Modal apps
|
|
8
|
+
- **Function Management**: Call and spawn Modal functions
|
|
9
|
+
- **Robot Inference**: Deploy robot inference clients on A100/H100 GPUs
|
|
10
|
+
- **Training Clients**: MjLab training on H100 GPUs
|
|
11
|
+
- **Video2Robot**: End-to-end video-to-policy training
|
|
12
|
+
- **Volume Management**: Create and manage persistent volumes
|
|
13
|
+
- **Secret Management**: Create and manage secrets
|
|
14
|
+
|
|
15
|
+
## Installation
|
|
16
|
+
|
|
17
|
+
```bash
|
|
18
|
+
bun add @ebowwa/mcp-modal
|
|
19
|
+
```
|
|
20
|
+
|
|
21
|
+
## Configuration
|
|
22
|
+
|
|
23
|
+
Set the following environment variables:
|
|
24
|
+
|
|
25
|
+
```bash
|
|
26
|
+
MODAL_TOKEN_ID=your-modal-token-id
|
|
27
|
+
MODAL_TOKEN_SECRET=your-modal-token-secret
|
|
28
|
+
```
|
|
29
|
+
|
|
30
|
+
## Usage with Claude Code
|
|
31
|
+
|
|
32
|
+
Add to your MCP configuration:
|
|
33
|
+
|
|
34
|
+
```json
|
|
35
|
+
{
|
|
36
|
+
"mcpServers": {
|
|
37
|
+
"modal": {
|
|
38
|
+
"command": "bun",
|
|
39
|
+
"args": ["run", "@ebowwa/mcp-modal"]
|
|
40
|
+
}
|
|
41
|
+
}
|
|
42
|
+
}
|
|
43
|
+
```
|
|
44
|
+
|
|
45
|
+
## Available Tools
|
|
46
|
+
|
|
47
|
+
### Health Check
|
|
48
|
+
- `modal_health` - Check Modal credentials and connection status
|
|
49
|
+
|
|
50
|
+
### App Management
|
|
51
|
+
- `modal_create_app` - Create or get a Modal app by name
|
|
52
|
+
|
|
53
|
+
### Function Management
|
|
54
|
+
- `modal_call_function` - Call a deployed Modal function
|
|
55
|
+
- `modal_spawn_function` - Spawn a function asynchronously
|
|
56
|
+
|
|
57
|
+
### Robot Inference
|
|
58
|
+
- `modal_deploy_robot_inference` - Deploy robot inference client (A100/H100)
|
|
59
|
+
- `modal_generate_action` - Generate robot action from observation
|
|
60
|
+
|
|
61
|
+
### Training
|
|
62
|
+
- `modal_deploy_mjlab_trainer` - Deploy MjLab trainer (H100)
|
|
63
|
+
- `modal_preprocess_motion` - Preprocess CSV motion data
|
|
64
|
+
- `modal_train_policy` - Train robot policy
|
|
65
|
+
|
|
66
|
+
### Video2Robot
|
|
67
|
+
- `modal_deploy_video2robot` - Deploy Video2Robot trainer
|
|
68
|
+
- `modal_process_video` - Process video to extract motion
|
|
69
|
+
- `modal_train_from_video` - Full pipeline: video -> policy
|
|
70
|
+
|
|
71
|
+
### Volume Management
|
|
72
|
+
- `modal_create_volume` - Create or get a persistent volume
|
|
73
|
+
|
|
74
|
+
### Secret Management
|
|
75
|
+
- `modal_get_secret` - Get a Modal secret
|
|
76
|
+
- `modal_create_secret` - Create a secret from key-value pairs
|
|
77
|
+
|
|
78
|
+
### Image Management
|
|
79
|
+
- `modal_image_from_registry` - Create image from registry tag
|
|
80
|
+
|
|
81
|
+
### Client Management
|
|
82
|
+
- `modal_list_deployed_clients` - List all deployed clients
|
|
83
|
+
- `modal_cleanup_client` - Clean up a deployed client
|
|
84
|
+
|
|
85
|
+
## GPU Types
|
|
86
|
+
|
|
87
|
+
| Type | Use Case |
|
|
88
|
+
|------|----------|
|
|
89
|
+
| T4 | Development, light workloads |
|
|
90
|
+
| A100 | Inference, moderate training |
|
|
91
|
+
| L4 | Cost-effective inference |
|
|
92
|
+
| H100 | Heavy training (mjlab) |
|
|
93
|
+
|
|
94
|
+
## License
|
|
95
|
+
|
|
96
|
+
MIT
|