Back to skill

Security audit

公众号文章阅读简报

Security checks for vulnerabilities and agentic risk

Overview

ClawScan’s benign verdict reflects no established malicious behavior and mostly purpose-aligned findings, with medium confidence because the scan was partial and flagged an unpinned external installer and a draft-editing capability.

The verdict is benign with medium confidence, not a guarantee of safety. Before installing, consider that the documented command uses an unpinned external installer and that optional draft editing is broader than article reading. The scan covered 12 of 13 components; one script was only partially inspected and binary 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied chunk is an installation command, not an implementation of the declared article-reading workflow. It targets an external skill repository rather than user-provided article URLs. The installed skill might implement the description, but its contents are not supplied and cannot be assessed here; no malicious behavior is established.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The only supplied executable behavior is external skill installation, which is not disclosed in the description. Although installation could be a setup step for a related skill, this chunk neither implements nor demonstrates the declared workflow. The installed skill's behavior cannot be determined from its repository name.

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 documented installation command invokes an unpinned npm CLI through npx, allowing the executable version to change between installations. Although this is not an MCP server reference, the underlying supply-chain risk is real: a compromised package release could execute code with the installer's local privileges; nothing here establishes that a compromise has occurred.

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 installation example uses npx to run the skills package without specifying a reviewed version, making installation dependent on mutable registry content. The MCP classification is inaccurate, but a compromised CLI release could still execute with the user's local privileges; the supplied source does not demonstrate an actual compromise.

Content

No source excerpt is available for this finding.

The reading-brief skill also supports modifying browser-based drafts, which is outside its manifest description.

Medium
Category
Excessive Agency
Confidence
92% confidence
Finding

The manifest describes reading specified public articles and producing a sourced brief, but L27 and L42 introduce an optional workflow that fills draft fields in the browser. Explicit authorization and the prohibition on publishing limit the impact, but draft modification remains a write capability not disclosed in the manifest's reading-and-summarization scope.

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.