Repository navigation
Expand file tree
/
Copy pathplugin.json
More file actions
112 lines (112 loc) · 6.58 KB
/
Copy pathplugin.json
File metadata and controls
112 lines (112 loc) · 6.58 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "worldlabs",
"version": "0.2.2",
"description": "Prepare images for 3D workflows with World Labs",
"extensions": {
"com.openai": {
"interface": {
"displayName": "World Labs",
"shortDescription": "Create new worlds with Atlas",
"developerName": "World Labs",
"category": "Creativity",
"composerIcon": "./assets/worldlabs-logo.png",
"logo": "./assets/worldlabs-logo.png",
"websiteURL": "https://www.worldlabs.ai",
"privacyPolicyURL": "https://www.worldlabs.ai/privacy-policy",
"termsOfServiceURL": "https://www.worldlabs.ai/terms-of-service",
"longDescription": "Turn photos into 3D-ready scenes with World Labs. Estimate the camera and depth of an image and download the results. Generate and edit scene images, create new views with Atlas, reconstruct Gaussian splats, and extract meshes. In Blender 5.3 or newer, use the Atlas extension to import calibrated projects, pose images, generate views, and build splats in your scene.",
"defaultPrompt": [
"Estimate the camera and depth for this photo of my room.",
"Show me which World Labs projects I can use.",
"Generate an image of a sunlit reading room."
],
"supportURL": "https://docs.worldlabs.ai/marble/support"
},
"review": {
"test_cases": {
"positive": [
{
"description": "Connect",
"prompt": "Show me which World Labs projects I can use.",
"expected_behavior": "Sign-in and consent complete in the browser; `list_projects` returns the account's projects with the task tools each can run in `available_tools`. No key is requested in chat.",
"tools_triggered": "list_projects"
},
{
"description": "Estimate camera and depth from an attached scene photo.",
"prompt": "Estimate the camera and depth for this photo.",
"expected_behavior": "`import_image`, then the project and credit use are named before `images2PosedRGBD`; `get_operation` polls to done and reports the camera, depth-grid size, and download links.",
"tools_triggered": "list_projects, import_image, images2PosedRGBD, get_operation"
},
{
"description": "Generate an image from a text prompt, with no attachment.",
"prompt": "Generate an image of a sunlit reading room.",
"expected_behavior": "The project and credit use are named before `atlasTextToImage`; `get_operation` polls to done and reports the image's download link.",
"tools_triggered": "list_projects, atlasTextToImage, get_operation"
},
{
"description": "After the generate case, estimate the camera and depth of the generated image in the same chat.",
"prompt": "Estimate the camera and depth for that image.",
"expected_behavior": "`images2PosedRGBD` names the generated image by its `assetId`; nothing is uploaded or imported again. `get_operation` polls to done and reports the camera and download links.",
"tools_triggered": "images2PosedRGBD, get_operation"
},
{
"description": "After the estimate case completes, check its returned operation ID and project ID in a new chat.",
"prompt": "(new chat) Check on operation `<id>` in project `<project>`.",
"expected_behavior": "`get_operation` returns the same operation; nothing is resubmitted.",
"tools_triggered": "get_operation"
},
{
"description": "After the estimate case, retry the same image request in the same chat with the original idempotency key.",
"prompt": "Retry that estimate with the same image and the same idempotency key. Do not start a new request.",
"expected_behavior": "images2PosedRGBD returns the original operation ID for the same project, image, and idempotency key; no duplicate operation is created.",
"tools_triggered": "images2PosedRGBD"
},
{
"description": "After the estimate case, request a download link for its depth asset again.",
"prompt": "Give me the depth file again.",
"expected_behavior": "get_asset returns a valid download link for the same depth asset. It may reuse a cached link while valid and renews it near expiry; the returned link downloads the depth file successfully.",
"tools_triggered": "get_asset"
}
],
"negative": [
{
"description": "A conceptual camera question requires no account access or image processing. Answer directly without invoking World Labs tools.",
"prompt": "Explain camera intrinsics versus extrinsics. Do not process any images or access my account.",
"expected_behavior": "Explain the concepts without calling World Labs tools."
},
{
"description": "A photography question about depth of field does not require estimating scene depth. Answer directly without invoking World Labs tools.",
"prompt": "How does aperture affect depth of field in a portrait?",
"expected_behavior": "Explain aperture and depth of field without calling World Labs tools."
},
{
"description": "The connected tools cannot change billing details. Explain the limitation without attempting unrelated tool calls.",
"prompt": "Change the billing address on my World Labs account.",
"expected_behavior": "Explain that billing-address changes are unsupported by the connected tools and direct the user to account settings or support; do not call World Labs tools."
}
]
}
},
"publication": {
"release_notes": "Initial release: connect World Labs to generate images and views, estimate cameras and depth, reconstruct Gaussian splats, extract meshes, and download results. Includes World Labs API guidance and Atlas for Blender Python workflows, with separate local tool and extension requirements."
}
}
},
"homepage": "https://www.worldlabs.ai",
"repository": "https://github.com/worldlabsai/skills",
"author": {
"name": "World Labs",
"url": "https://www.worldlabs.ai",
"email": "support@worldlabs.ai"
},
"license": "Apache-2.0",
"keywords": [
"worldlabs",
"atlas",
"blender",
"3d",
"images",
"gaussian-splat"
]
}