ClawHub
Back to skill

Security audit

OpenClaw Web Clipper

Security checks across malware telemetry and agentic risk

Overview

This skill does what it claims: it clips user-provided web pages into local Markdown files, with ordinary local-file and network risks for that purpose.

Install only if you want your agent to fetch requested web pages and store their contents locally. Keep the output directory set to a dedicated clippings folder, review batch URL lists before running them, and be cautious clipping private or paywalled pages because their contents may become searchable in your local knowledge base.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Taint TrackingDirect Taint Flow, Variable-Mediated Taint Flow, Credential Exfiltration Chain
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Tainted flow: 'fpath' from os.environ.get (line 200, credential/environment) → open (file write)

Medium
Category
Data Flow
Content
*由 xiaoyaoclaw-web-clipper 自动剪藏*
"""

    with open(fpath, "w", encoding="utf-8") as f:
        f.write(md)

    # 更新索引
Confidence
73% confidence
Finding
The output path is influenced by CLIPPER_OUTPUT_DIR or the --dir argument and is used directly for file creation, allowing writes to arbitrary filesystem locations accessible to the process. In a local CLI tool this is often expected behavior, but in an agent or automated skill context it becomes more dangerous because external orchestration or environment manipulation could redirect writes into sensitive directories, overwrite operational files, or plant content for later processing.

Lp3

Medium
Category
MCP Least Privilege
Confidence
97% confidence
Finding
The skill instructs the agent to perform network access, write Markdown and index files locally, and read environment-derived configuration such as CLIPPER_OUTPUT_DIR, but it does not declare those permissions explicitly. This creates a transparency and policy-enforcement gap: a host system or user may assume the skill is low-risk while it can fetch arbitrary URLs and persist content to disk, which can enable unintended data ingestion, storage abuse, or writes in sensitive locations if the runtime honors the skill's instructions.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The README suggests very generic trigger phrases such as '保存这篇文章' and '收藏这个网页', which can overlap with ordinary conversation and increase the chance that an agent invokes the skill unintentionally. In the context of a web clipper, accidental invocation can cause the agent to fetch remote content and write local files without the user realizing that a network and filesystem action is being taken.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The README describes convenience and usage but does not prominently warn that using the skill will retrieve data from external URLs and persist content into local files under knowledge/clippings/. That omission matters because users may treat the action as a harmless summarization step, when it actually performs network access and local writes that could ingest malicious or sensitive content into the workspace.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger phrases are broad and map to common requests like saving or collecting a page, increasing the chance the skill will auto-activate when the user did not intend local network fetching and file writes. In this skill's context that matters because invocation causes external content retrieval and persistent filesystem changes, so accidental activation can lead to unwanted downloads, storage of sensitive pages, or operational confusion.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.