Back to skill

Security audit

Maxhub Sora2

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report capabilities and fallback behavior that may go beyond its stated Sora2 data-query purpose; this is a review warning, not a claim of malicious intent.

Review the upload and fallback capabilities and decide whether they match your needs before installing. The install command does not pin a package version, and the Skillspector scan was partial: three local references were unresolved, so the evidence is incomplete.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

MCP server referenced without pinned version: 'npx clawhub'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The installation command executes an unpinned npm package through npx, making installation dependent on whichever release resolves at that time. A compromised future release could execute code with the installing user's permissions; this is a supply-chain exposure, not specifically an MCP-server issue.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx clawhub'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The documented npx command does not pin the executable package version, allowing installation behavior to change without a documentation update. If the resolved release is compromised, running the command can execute attacker-controlled code with the user's permissions; no MCP server is actually identified here.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

md
maxhub_auth_header="Authorization: Bearer ${MAXHUB_API_KEY}"

# GET example
curl -s "https://www.aconfig.cn/api/v1/sora2/{endpoint}?%7Bparams=}" \
  -H "$maxhub_auth_header"

# POST example

The Cameo router exposes image uploading despite explicit read-only guarantees.

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

The skill categorically promises read-only operation while routing Cameo requests to an image-upload capability and directing autonomous execution of planned calls. This mismatch could cause users or agents to permit external image transmission under a mistaken read-only assumption, although the supplied instructions do not demonstrate an actual unauthorized upload.

Content

No source excerpt is available for this finding.

The fallback workflow directs the agent to Douyin APIs outside the declared Sora2 scope.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The automatic fallback table supplies Douyin endpoints despite the skill's declared Sora2 purpose. If selected, these substitutions could send query identifiers to an unintended platform-specific service, consume API quota, or return unrelated data as a successful result; the destination host remains the declared provider.

Content

No source excerpt is available for this finding.

The skill exposes an image-upload operation despite being described as a Sora2 data-query assistant.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The manifest describes retrieving platform data, but this reference exposes POST /api/v1/sora2/upload_image to upload images for subsequent AI video generation. The documentation explicitly states that uploaded images are stored on the server, making this a remote write capability rather than an implementation detail of querying Cameo or user data.

Content

No source excerpt is available for this finding.

Example prompts use generic words without requiring Sora2-specific intent.

Low
Category
Trigger Abuse
Confidence
95% confidence
Finding

The example prompts include 用户 (user), 上传 (upload), and 图片 (image), which commonly occur in requests unrelated to Sora2. Although the README identifies the platform, it does not explicitly constrain these generic prompts to Sora2 requests or provide exclusions, leaving the invocation scope ambiguous.

Content

No source excerpt is available for this finding.

Invocation examples use broad, context-free words such as “用户” (user), “上传” (upload), and “图片” (image).

Low
Category
Trigger Abuse
Confidence
94% confidence
Finding

The example instructions at L28–L30 consist of isolated words that commonly occur in requests unrelated to Sora2. No activation constraints or exclusion examples clarify that these words should invoke the skill only for Sora2-related tasks.

Content

No source excerpt is available for this finding.