Back to skill

Security audit

Maxhub Bilibili

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it flags an out-of-scope Douyin fallback and an unpinned installer command; the available scan is partly incomplete.

Consider whether the Douyin fallback fits your needs and whether you trust an unpinned installer. The scanner inspected all listed components but marked analysis partial because some references were missing; this evidence is incomplete, so the verdict does not establish that the skill is safe.

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 (6)

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

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

The installation example executes an unpinned npm CLI through npx, allowing the resolved package version to change without review. A compromised future release could execute code with the installer's privileges; this is a supply-chain risk even though clawhub is an installer rather than an MCP server.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

The documented npx invocation does not pin the clawhub package version, so installation can execute a release that has not been reviewed. If that release is compromised, it could run arbitrary code with the user's privileges; the risk is unpinned installer execution, not an MCP-specific issue.

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/bilibili/{endpoint}?%7Bparams=}" \
  -H "$maxhub_auth_header"

# POST example

The Bilibili-only skill directs fallback requests to Douyin APIs.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The fallback table directs a Bilibili assistant to make authenticated Douyin requests automatically after failures, despite the different platform scope. This can submit query parameters to an unintended API operation, consume paid calls, or produce misleading cross-platform results; the unchanged provider hostname limits the evidence for additional third-party disclosure.

Content

No source excerpt is available for this finding.

The example invocation prompts include generic words such as “搜索” (search), “搜” (search), “回复” (reply), and “推荐” (recommend), without requiring Bilibili-specific context.

Low
Category
Trigger Abuse
Confidence
95% confidence
Finding

These standalone words overlap with ordinary requests unrelated to Bilibili and could cause unintended skill invocation. Although the README identifies the skill as a Bilibili assistant, it does not explicitly restrict these example prompts to Bilibili-related requests or provide exclusion conditions.

Content

No source excerpt is available for this finding.

Example invocation phrases are overly broad and do not require Bilibili-specific context.

Low
Category
Trigger Abuse
Confidence
92% confidence
Finding

The example commands include generic terms such as “搜索” (search), “回复” (reply), and “推荐” (recommend), which commonly occur in requests unrelated to Bilibili. Although the introduction identifies the skill's domain, the invocation examples do not explicitly require that context or exclude unrelated requests.

Content

No source excerpt is available for this finding.