Back to skill

Security audit

ArchLang

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious with medium confidence, mainly because it recommends running unpinned npm packages that could change without review; this is a supply-chain risk, not evidence of malware.

Consider the supply-chain risk before installing. Review the package and installer sources, and prefer exact, reviewed versions. Treat external playground links as potentially exposing plan content unless their sharing behavior is clear. The scan covered one Markdown component, but reference resolution 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
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (6)

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

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The compatibility instructions recommend executing an npm package without a fixed version, allowing the code selected on first installation or a later resolution to change without review. This is a supply-chain exposure rather than evidence of malware: a compromised release could execute with the agent's filesystem and environment access. The scanner's MCP label is inaccurate here because this invocation runs the CLI.

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
95% confidence
Finding

The zero-install setup executes an unpinned npm package through npx, and -y suppresses its installation confirmation. If the resolved release or its dependencies are compromised, even the apparently harmless --help invocation can execute attacker-controlled code. This repeats the same supply-chain issue as the compatibility instructions, not a separate exploit mechanism.

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
95% confidence
Finding

The instructions generalize the unpinned npx command to every subsequent CLI operation and also offer an unpinned global installation. This makes execution of an unreviewed release part of the normal workflow, although caching means it does not necessarily fetch new code on every call. This is another occurrence of the same dependency-trust weakness.

Content

No source excerpt is available for this finding.

The delivery workflow instructs the agent to create an external playground link without a privacy warning about sharing floor-plan content.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The default delivery workflow directs the agent to produce an external playground URL that opens the user's plan without first explaining the sharing implications or obtaining consent. Floor plans can disclose private layouts and access routes; if the URL embeds the source or references a hosted copy, opening or sharing it can expose that information to recipients and potentially the playground service. The supplied documentation does not establish whether link generation uploads data or encodes it locally, so immediate network exfiltration is not proven.

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
94% confidence
Finding

The delivery example again permits execution of the CLI from an unpinned npm release. A compromised release could access the floor-plan file and other resources available to the agent when generating the link. The link operation itself does not prove malicious behavior, and this occurrence shares the earlier supply-chain root cause.

Content

No source excerpt is available for this finding.

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

Medium
Category
MCP Rug Pull
Confidence
87% confidence
Finding

The optional npx skills add command executes an unpinned installer and references a skill repository without a specified immutable revision. Compromise of either source could introduce executable code or adversarial agent instructions. Its placement as an optional discovery note reduces exposure compared with the mandatory setup steps, and it is not an MCP-server invocation.

Content

No source excerpt is available for this finding.