Back to skill

Security audit

Reverse-Engineering Assistant

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a clearly scoped reverse-engineering workflow, with expected risks from downloading tools and running analyzed software.

The CLI fallback may download and execute third-party code without another npm prompt. Use a restricted analysis directory, retain the version pin, and review updates and setup changes. Running target software is not sandboxed; approve it only after considering its file and network access.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Automatic CLI fallback downloads and executes third-party code without warning the user.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The fallback instructs the agent to run npx -y rea-agents@6.5.0 <command>, which can download and execute third-party code with the agent's permissions while suppressing npm's installation prompt. Although the version is pinned and the package serves the skill's stated purpose, no explicit approval or package-integrity verification is required; a compromised package or dependency could access local files, alter the workspace, or exfiltrate analysis targets. The confirmation rules cover running target software and exporting files, but do not clearly cover installing and executing the analysis tool itself.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx rea-agents'.

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The example command npx rea-agents@latest setup uses a mutable version tag, so following it may download and execute a newly published package version without prior review. A compromised package release could execute code with the invoking user's permissions and affect accessible projects or evidence files. This is an update-record template rather than an automatic execution instruction, which reduces immediate exposure; the excerpt also does not establish that this package is an MCP server.

Content

No source excerpt is available for this finding.