Back to skill

Security audit

Acp Adapter Layer

Security checks for vulnerabilities and agentic risk

Overview

This skill appears to be a legitimate IDE-to-agent bridge, but it needs review because it can pass prompts and session history to a gateway while its implementation lacks strong validation and session isolation.

Review before installing or using with confidential workspaces. Use it only with a trusted local OpenClaw gateway, and require stronger input validation, cryptographically secure session IDs, caller authorization, and clear session-history retention/deletion controls before production use.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
index.ts:76
Finding

ACP Request Parameters Are Not Validated

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
index.ts:103
Finding

Predictable Session Identifiers Without Client Authorization Binding

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly describes forwarding ACP prompts, session state, and session history through an OpenClaw gateway, including session/history retrieval and WebSocket transport, but it does not warn users that potentially sensitive prompts and histories leave the immediate IDE context. In an IDE integration skill, this omission can cause inadvertent disclosure of source code, secrets, or proprietary context because users may assume interactions remain local unless clearly told otherwise.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
72% confidence
Finding

The skill design includes creation of factory sessions and loading/replaying ASF session history, indicating persistent session state across ACP tasks without any documented retention limits, consent flow, or data minimization guarantees. In a multi-agent IDE bridge, persistent sessions can accumulate sensitive code, prompts, and tool outputs, increasing exposure if accessed by other components, replayed unexpectedly, or retained longer than intended.

Content

Scanner excerpt · SKILL.md (reported line 596)May include surrounding context.

typescript
interface AgenticFactoryIntegration {
  // Create factory session for ACP task
  createFactoryForACP(acpSession: ACPSession): Promise<FactorySession>;
  
  // Route ACP prompt to factory

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The inline comment says the real behavior would route the prompt to an ASF agent, indicating actual forwarding/integration behavior. However, the code below does not perform any routing or gateway use and instead returns a hardcoded placeholder response, which directly contradicts the documented intent of this handler.

Content

No source excerpt is available for this finding.