Back to skill

Security audit

xhs-image-note-release

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because scanners report instructions to disable sandbox isolation and a browser-rendering risk, while some artifact analysis was incomplete.

Consider avoiding installation unless you can review the incomplete script analysis and accept the reported sandbox and browser-rendering risks. The verdict is suspicious, not a finding that the author is malicious; scanner coverage was partial, 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
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
SKILL.md
Finding

Instructions prescribe disabling agent isolation after sandbox rejection

Content
View full analysis

Impact

Following these instructions removes the isolation that blocked execution and expands access available to subsequent agent commands. No subsequent credential theft or host compromise was established.

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

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 127)May include surrounding context.

md
> 本 skill 的 `scripts/publish_note.mjs` 是 **exit 60 中止**。两套脚本各自独立,行为不同是有意的。

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 316)May include surrounding context.

md
- `scripts/publish_note.mjs` — 流程层,所有浏览器操作,可用 `node --check` 单独校验

Instructions to disable WorkBuddy's sandbox omit a warning about reduced security isolation.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The README directs WorkBuddy users to disable the agent sandbox as a compatibility workaround, without explaining the loss of isolation or limiting the change. Following this advice could expose host files and other resources to subsequent agent commands or compromised dependencies; the authenticated browser workflow makes that exposure more consequential.

Content

No source excerpt is available for this finding.

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

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.

Generic draft-saving triggers can activate the Xiaohongshu publishing skill outside its intended context.

Medium
Category
Trigger Abuse
Confidence
90% confidence
Finding

The trigger list includes “存草稿” (save a draft) and “推到草稿箱” (move to drafts) without requiring a Xiaohongshu context. These phrases also apply to email, documents, and other publishing platforms, and the skill gives no explicit exclusion or contextual requirement for them.

Content

No source excerpt is available for this finding.

Instructions to disable the WorkBuddy sandbox omit a warning about the loss of security isolation.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The skill repeatedly instructs users to disable WorkBuddy's sandbox to resolve browser interruptions, without warning about reduced host isolation or requiring its restoration. This broad security downgrade increases the potential reach of malicious inputs or compromised scripts during a workflow that processes local files and uses an authenticated browser.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
96% confidence
Finding

The subprocess invocation uses an argument list without a shell, so it does not demonstrate shell-command injection. However, it launches Chrome with --no-sandbox to render HTML containing unsanitized SVG illustration content, allowing attacker-supplied active content into an unsandboxed renderer. A browser renderer exploit could consequently compromise the host with the invoking user's privileges.

Content

Scanner excerpt · references/card-generator/card_generator.py (reported line 1292)May include surrounding context.

python
"file://" + tmp_html.name,
        ]

        result = subprocess.run(cmd, capture_output=True, text=True)

        if result.returncode != 0:
            print(f"[ERROR] 渲染失败: {result.stderr}", file=sys.stderr)

The example workflow proceeds from card generation to publishing without warning the user about external visibility.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Step 3 instructs the agent to publish using ego-browser, although the example user request only asks for photo-background quote cards. This file does not warn that the generated content will be published or require approval before making it externally visible.

Content

No source excerpt is available for this finding.

The document instructs users to disable the WorkBuddy sandbox without warning about the loss of security isolation.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The document explicitly requires disabling the WorkBuddy sandbox and repeats that recommendation as troubleshooting guidance, removing a security boundary rather than granting narrowly scoped browser access. In an agent workflow that handles external content and an authenticated browser, this increases the consequences of prompt injection or software exploitation by exposing additional host resources. Its historical-reference disclaimer reduces routine exposure but does not make the advice safe.

Content

No source excerpt is available for this finding.