Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

tripo3d

v1.0.1

Tripo3D AI 3D model generation. Use when generating 3D models from text prompts or images via the Tripo3D API. Supports Text-to-3D, Image-to-3D, Multiview-to...

1· 88·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for richardcoder849/tripo3d.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "tripo3d" (richardcoder849/tripo3d) from ClawHub.
Skill page: https://clawhub.ai/richardcoder849/tripo3d
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install tripo3d

ClawHub CLI

Package manager switcher

npx clawhub@latest install tripo3d
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
The SKILL.md clearly requires a TRIPO3D_API_KEY and describes calling the Tripo3D API (endpoints, uploads, downloads), which is coherent with the skill's name and description. However, the registry metadata lists no required env vars or primary credential — that is inconsistent and misleading for users who expect credential requirements to be declared. The SKILL.md also assumes PowerShell/.NET for downloads (Windows-specific) but the skill declares no OS or binary requirements.
Instruction Scope
Instructions focus on creating tasks, uploading images, polling status, and downloading models — all within the stated purpose. Concerns: (1) the doc strongly pushes PowerShell/.NET WebClient and a particular local proxy (Clash Verge on port 7897), which is operational guidance that could route traffic through a local proxy; (2) polling guidance is inconsistent (it says poll every 5–10s, but also says 'Never block with sleep' and 'poll only when user asks'); (3) SKILL.md requires access to an API key (not declared in metadata). The instructions do not ask the agent to read unrelated files or other credentials.
Install Mechanism
No install spec and no code files — instruction-only skill. This minimizes on-disk footprint and is the lowest-risk install pattern.
!
Credentials
The runtime docs require TRIPO3D_API_KEY (appropriate for an API client), but the registry metadata did not declare any required env vars or a primary credential. That discrepancy is a red flag: users may not realize they must supply an API key. No other credentials are requested, which is proportionate to the described functionality.
Persistence & Privilege
The skill is not always-enabled and does not request elevated or persistent platform privileges. It is user-invocable and allows autonomous invocation (the platform default); that alone is expected and not a reason to flag.
What to consider before installing
The SKILL.md appears to implement a legitimate Tripo3D API client, but the registry metadata does not declare the TRIPO3D_API_KEY that the instructions require — this mismatch is the main concern. Before installing: (1) confirm the skill owner and whether the registry metadata will be corrected to list TRIPO3D_API_KEY; (2) only provide an API key you can revoke (create a scoped/test key if the platform allows); (3) be aware the instructions favor PowerShell/.NET and a local proxy (127.0.0.1:7897) — routing traffic through a proxy can expose requests to interception, so avoid enabling unknown system proxies; (4) test on an isolated account/environment first to confirm behavior; (5) ask the maintainer why the metadata omits required env vars and why Python/aiohttp is discouraged. These checks will reduce risk and clarify whether the inconsistencies are just sloppy packaging or something more suspicious.

Like a lobster shell, security has layers — review code before you run it.

latestvk9724tbe6eksdydjspvptsjmh9843gqa
88downloads
1stars
1versions
Updated 3w ago
v1.0.1
MIT-0

Tripo3D AI 3D Generation Skill

Overview

Tripo3D is a leading AI 3D generation platform. This skill handles:

  • Text → 3D: Natural language prompt to 3D model
  • Image → 3D: Single image to 3D model (photo to 3D)
  • Multiview → 3D: 4 images (front/left/back/right) to higher fidelity 3D
  • Refine Model: Enhance an existing draft model
  • Polling: Check generation status and retrieve results
  • Download: GLB/FBX model files + preview images

Setup

Requires TRIPO3D_API_KEY environment variable.

Get your API key at: https://platform.tripo3d.ai

Proxy Configuration (Windows + Clash Verge)

Clash Verge (verge-mihomo) proxy port: 7897 — not 7890.

Use PowerShell for all API calls:

Invoke-RestMethod -Uri '...' -Proxy 'http://127.0.0.1:7897' -ProxyUseDefaultCredentials

For file downloads, use .NET WebClient:

$wc = New-Object System.Net.WebClient
$wc.Proxy = New-Object System.Net.WebProxy("http://127.0.0.1:7897")
$wc.Proxy.Credentials = [System.Net.CredentialCache]::DefaultCredentials
$wc.DownloadFile($url, $outPath)

Python SDK (aiohttp) does NOT respect system proxy — use PowerShell instead.

Base URL

https://api.tripo3d.ai/v2/openapi

All requests require header:

Authorization: Bearer <TRIPO3D_API_KEY>
Content-Type: application/json

Core Endpoints

ActionMethodPath
Create taskPOST/task
Check statusGET/task/{task_id}
Upload image (STS)POST/upload/sts
Get balanceGET/user/balance

Task Types

All generation uses POST /task with different type field values:

TypeDescription
text_to_modelText prompt → 3D model
image_to_modelSingle image → 3D model
multiview_to_model4 images (front/left/back/right) → 3D model
refine_modelRefine an existing draft model

Workflow 1: Text to 3D

Step 1 — Create task

POST /v2/openapi/task
{
  "type": "text_to_model",
  "prompt": "a fierce tiger with orange and black stripes"
}

Step 2 — Poll status

GET /v2/openapi/task/{task_id}

Poll every 5–10 seconds. Done when status === "success".

Step 3 — Download model From response: data.output.pbr_model.url


Workflow 2: Image to 3D

Step 1 — Upload image (STS)

POST /v2/openapi/upload/sts
Content-Type: multipart/form-data

Form field: file (JPEG/PNG/WEBP, max 10MB, min 256px)

Response:

{ "code": 0, "data": { "image_token": "uuid-xxx" } }

Step 2 — Create image-to-3D task

POST /v2/openapi/task
{
  "type": "image_to_model",
  "file": {
    "type": "jpg",
    "file_token": "<image_token>"
  }
}

Or use image URL directly (no upload needed):

{
  "type": "image_to_model",
  "file": {
    "type": "jpg",
    "url": "https://example.com/photo.jpg"
  }
}

Step 3 — Poll status (same as Text-to-3D)

Step 4 — Download model


Workflow 3: Multiview to 3D

Upload 4 images (front, left, back, right), then:

POST /v2/openapi/task
{
  "type": "multiview_to_model",
  "files": [
    { "type": "jpg", "file_token": "<front_token>" },
    { "type": "jpg", "file_token": "<left_token>" },
    { "type": "jpg", "file_token": "<back_token>" },
    { "type": "jpg", "file_token": "<right_token>" }
  ]
}

Front image is required; other 3 can be empty {}.


Workflow 4: Refine Model

POST /v2/openapi/task
{
  "type": "refine_model",
  "draft_model_task_id": "<task_id_of_draft>"
}

Note: Only works with model_version < v2.0-20240919 drafts.


Common Parameters

ParameterApplies toDescription
model_versionallVersion: P1-20260311, Turbo-v1.0-20250506, v3.1-20260211, v3.0-20250812, v2.5-20250123 (default), v2.0-20240919, v1.4-20240625
textureallEnable texture, default true
pbrallEnable PBR, default true
texture_qualityall"detailed" or "standard" (default)
face_limitallLimit polygon count (1000~20000)
quadtext/imagetrue → FBX output (not GLB)
compressall"geometry" for compressed output
auto_sizeallScale to real-world meters
texture_alignmentimage/multiview"original_image" or "geometry"
model_seedallInteger, reproducible generation
texture_seedallInteger, reproducible texture
enable_image_autofiximage/multiviewAuto-enhance input image
generate_partstext/imageSegmentation mode

P1-20260311 (Low-poly optimized)

Use for game assets, stylized/low-poly models. Only supports: model_version, model_seed, face_limit (48~20000), texture, pbr, texture_seed, texture_quality, auto_size, compress, export_uv.


Polling Logic

⚠️ Generation is asynchronous. Poll every 5–10 seconds.

Status values:

  • queued → waiting in queue
  • running → actively generating
  • success → done, use output
  • failed / banned / expired / cancelled → check error

Never block with sleep. Submit job, reply with task_id, poll only when user asks.


Download

Model URLs expire after ~5 minutes. Download immediately after success.

Use .NET WebClient (PowerShell):

$wc = New-Object System.Net.WebClient
$wc.Proxy = New-Object System.Net.WebProxy("http://127.0.0.1:7897")
$wc.Proxy.Credentials = [System.Net.CredentialCache]::DefaultCredentials
$wc.DownloadFile($modelUrl, $outPath)

Output files from data.output:

  • pbr_model.url → GLB file (main model)
  • generated_image.url → AI concept preview image
  • rendered_image.url → Model rendering preview

Cost

TaskCredits
Text-to-3D~20
Image-to-3D~20
Multiview-to-3D~20
Refine Model~20

Check balance: GET /v2/openapi/user/balance


Error Handling

CodeMeaning
2010Insufficient credit
2001Task not found (wrong task_id or API key mismatch)
2002Unsupported task type
2003Empty input file
2004Unsupported file type
429 2000Rate limited — retry later

Windows PowerShell Upload Example

$headers = @{
  'Authorization' = 'Bearer <API_KEY>'
}
$filePath = "C:/path/to/image.jpg"
$response = Invoke-RestMethod -Uri 'https://api.tripo3d.ai/v2/openapi/upload/sts' `
  -Method POST `
  -Headers $headers `
  -ContentType 'multipart/form-data' `
  -Proxy 'http://127.0.0.1:7897' `
  -ProxyUseDefaultCredentials `
  -Body @{
    file = Get-Item $filePath
  }
$imageToken = $response.data.image_token

Comments

Loading comments...