Back to skill

Security audit

multi-agent-comm

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent multi-agent communication guide, but it needs review because it encourages persistent agents, session resumption, and environment/API-key sharing without enough safety guidance.

Review this before installing if your agents have access to email, documents, calendars, API keys, or prior session history. Use only trusted subagents, avoid broad environment inheritance where possible, verify any resumed session belongs to the intended user and task, and rotate or provision per-device API keys rather than copying secrets casually.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill explicitly states that child agents inherit parent process environment variables, but provides no warning about secrets such as API keys, tokens, or credentials being exposed to subagents or external ACP-configured agents. In a multi-agent communication framework, this increases the risk of unintended secret propagation across trust boundaries, especially when spawning independent or differently configured agents.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The guide explicitly tells users to resume a session using an identifier sourced from a local session directory, but it does not warn that doing so may attach to an existing session containing prior prompts, secrets, or privileged context. In a multi-agent communication skill, reattaching to the wrong session can leak historical conversation data across agents or users and can cause actions to execute under unintended context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The migration instructions tell users to move skills to a new device and ensure API keys are present, but they omit any guidance on secure credential transfer, storage, rotation, or least-privilege setup. In practice, this can lead users to copy secrets insecurely, leave keys in shell history or plaintext files, or replicate broad production credentials onto less trusted devices.

Content

No source excerpt is available for this finding.