Back to skill

Security audit

ship-skill

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because it reports instructions to override telemetry opt-outs, bypass a deletion safety block, and repeatedly run unpinned command line tools; some script analysis was incomplete.

Consider the verdict’s suspicious Review status before installing. The reported telemetry override and safety-block workaround deserve particular attention, and unpinned commands can run changing packages. Scanner coverage was partial: one script was not fully interpreted, so its behavior remains uncertain.

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
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
references/pitfalls.md
Finding

Alternative forced deletion recommended after a safety-policy block

Content
View full analysis

Impact

An agent following the alternative remedy could delete the blocked test directory outside the original safety mechanism. The evidence does not establish broader deletion targets or that the alternative command would succeed in every environment.

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (35)

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding

Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Content

No source excerpt is available for this finding.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 126)May include surrounding context.

md
node scripts/check.mjs <owner>/<repo> [skill-name]

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 172)May include surrounding context.

md
- `scripts/check.mjs`:发布状态自检

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: network.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

This is a CLI invocation, not an MCP server, but the underlying supply-chain concern is valid: unversioned npx can download and execute a changing package release. A compromised release could execute with the agent's filesystem access and environment privileges, even though the requested operation only lists skills.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

The unversioned clawhub login command can execute a newly resolved package release during an authentication workflow. Although this is not an MCP server, a compromised CLI could interfere with authentication or access credentials available to the process.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
97% confidence
Finding

Publishing through unversioned npx executes a mutable dependency with access to repository contents and potentially authenticated publishing credentials. A compromised package release could steal credentials or publish modified content; the MCP-specific label is inaccurate, but the supply-chain risk remains.

Content

No source excerpt is available for this finding.

Reframes existing telemetry opt-outs as an unclean environment and instructs the agent to override them.

Medium
Category
Prompt Injection
Confidence
99% confidence
Finding

The instructions characterize telemetry opt-outs as an unclean environment and tell the agent to clear them, treating privacy controls as obstacles to successful publication. This is the same underlying issue as SQP-2 and could cause an agent to disregard user or organizational privacy preferences without a separate authorization decision.

Content

No source excerpt is available for this finding.

The workflow overrides telemetry opt-outs without warning the user about the privacy impact.

Medium
Category
Output Handling
Confidence
99% confidence
Finding

The command explicitly clears DISABLE_TELEMETRY and DO_NOT_TRACK for the installation process, enabling telemetry despite existing opt-outs. Discoverability is a legitimate goal, but it does not establish consent to disclose installation activity or associated identifiers; the text itself notes machine identification.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

This command executes an unpinned skills CLI and installs content with automatic confirmation. A compromised CLI release could run with the agent's privileges; the separate telemetry override adds privacy risk but is not the cause of this supply-chain finding.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% 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 136)May include surrounding context.

curl -s "https://www.skills.sh/api/search?q="

clawhub 下载量

curl -s "https://api.clawhub.ai/api/search?q="

text

CLI 能装、页面 200、README 有链接,都不能证明被索引。只认 API 返回的 `count`。

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

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The login instructions execute an unpinned npm package through npx. Although this is a CLI rather than an MCP server, a compromised future release could execute with the user's permissions during an authentication workflow and access credentials or repository contents.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The identity-check command still executes an unpinned npm CLI; a read-only subcommand does not make package execution read-only. If the resolved package release is compromised, it could access the authenticated user's token and local files.

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 publish command executes an unpinned CLI in a context containing publishing credentials and release contents. A compromised resolved release could steal those credentials or alter what is published; the publishing purpose makes this exposure more consequential.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The layout-validation example runs an unpinned npm CLI. The --list option limits intended functionality, not the permissions of downloaded code, so a compromised release could still read or modify local files.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
93% confidence
Finding

The installation example executes an unpinned npm package and automatically accepts installation prompts with -y. A compromised CLI release could execute arbitrary code with the user's permissions; automatic acceptance also removes an opportunity to inspect installation choices.

Content

No source excerpt is available for this finding.