Back to skill

Security audit

公众号素材库目录

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill benign with medium confidence: its declared read-only materials review does not align cleanly with scanner-reported installation and extra capabilities, and some content was only partially analyzed.

The verdict is benign, not risk-free. Before installing, consider the unpinned installer and the reported mismatch between the stated purpose and additional capabilities. Analysis was partial: 13 of 14 components were scanned, one file was only partly inspected, and two images were outside the analyzer’s scope.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (7)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk is an installation command, not an implementation of the declared inventory workflow. Adding an external skill is a materially different operation from read-only inspection of visible materials. The contents and behavior of the downloaded skill are not supplied and cannot be assumed to match the description.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The executable behavior supplied is skill installation, which can retrieve external content and modify the local skill environment. That is distinct from the declared one-time, read-only review of authorized browser content. Although the repository name is relevant to the stated task, it does not establish the installed skill's behavior or safeguards; those cannot be verified from this command.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The installation command invokes an unpinned npm CLI, although the finding incorrectly describes it as an MCP server. On a fresh installation, npx can retrieve and execute a changing package release with the user's privileges; a compromised release could therefore affect the local system despite the skill's read-only browser restrictions.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The documented npx command does not pin the skills CLI version; this is an npm execution risk rather than an MCP-server configuration issue. A compromised package release selected during installation could execute locally with the user's privileges, outside the browser workflow's read-only controls.

Content

No source excerpt is available for this finding.

The skill permits filling website drafts despite its read-only manifest.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

The manifest at L03 limits the skill to read-only inspection and delivery of a Markdown draft. However, L26 introduces draft_fill_authorized and L61 permits filling a website draft, which modifies remote content; requiring authorization and avoiding publication do not make that operation read-only.

Content

No source excerpt is available for this finding.

Scope Creep

Low
Category
Excessive Agency
Confidence
70% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · LICENSE (reported line 7)May include surrounding context.

text
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

The script implements promotional-image rendering, which is outside the declared browser-based inventory and Markdown-draft purpose.

Low
Category
Excessive Agency
Confidence
92% confidence
Finding

The script loads a local image, overlays promotional text on a 1600×900 canvas, and writes a PNG to a command-line-specified output path. This image-generation capability is not needed to inspect visible WeChat materials or produce the promised source-linked Markdown inventory, and the manifest does not declare promotional artwork as an output.

Content

No source excerpt is available for this finding.