Back to skill

Security audit

ArchLang

Security checks for vulnerabilities and agentic risk

Overview

The scanners classify this skill as benign, with medium confidence, while flagging unpinned package commands and a possible privacy issue around playground links; coverage is partial.

The verdict is benign, not risk-free. Before installing, consider whether you are comfortable with unpinned package commands and with a workflow that may create a playground link containing floor-plan information. Coverage is incomplete: the scanner reported unresolved local references, so its results do not establish that every referenced item was reviewed.

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
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

MCP server referenced without pinned version: 'npx -y @chanmeng666/archlang'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The compatibility instructions recommend executing an npm package without specifying a version, allowing the code obtained during setup to change without review. Although this is a CLI rather than an MCP server, a compromised package release or dependency could execute with the agent's filesystem and credential access.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @chanmeng666/archlang'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The setup command uses unversioned npx execution, and -y suppresses the installation confirmation. This is a conventional setup pattern rather than evidence of malicious intent, but it can automatically execute a newly compromised release with the agent's privileges.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @chanmeng666/archlang'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The instruction to substitute unversioned npx calls for every CLI invocation propagates the same supply-chain exposure throughout the workflow. It makes the executable version environment-dependent rather than tying execution to reviewed code; it does not prove that every invocation downloads a fresh release.

Content

No source excerpt is available for this finding.

The delivery workflow instructs generating a playground link for the user's floor plan without a privacy warning.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The standard delivery workflow directs the agent to generate a third-party playground link without first addressing whether the user permits sharing their building layout. The supplied text does not establish whether link generation uploads data or encodes it locally, but a plan-bearing URL can disclose sensitive layout information when shared or opened, depending on the URL format and service behavior.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @chanmeng666/archlang'.

Medium
Category
MCP Rug Pull
Confidence
98% confidence
Finding

The delivery step again recommends executing an unpinned npm package, this time while processing the user's completed floor plan. A compromised release could access that plan and other files available to the agent; this repeats the setup vulnerability rather than establishing a separate malicious mechanism.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
95% confidence
Finding

The optional npx skills command executes an unversioned installer and references a repository without an immutable revision. Although neither is an MCP server, a compromised installer could execute code, and changed repository content could introduce unreviewed instructions into the agent's future workflow.

Content

No source excerpt is available for this finding.