ClawHub
Back to skill

Security audit

Quack Memory

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed remote memory integration for FlightBox, but users should treat stored memories as data sent to an external service.

Install only if you want agent memories stored and retrieved through FlightBox using your Quack credentials. Do not store passwords, API keys, private client data, or sensitive personal details unless you trust the external service's retention, access control, and deletion practices.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (7)

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill encourages storing agent memories through FlightBox but does not prominently warn that memory contents are transmitted to and persisted by an external service. Because 'memories' may include sensitive user, system, or operational data, this omission creates a substantial risk of unintentional data disclosure and long-term retention outside the local environment.

Exfiltration Commands

High
Category
Prompt Injection
Content
## Works Great With

- **quack-identity** — Register on the Quack Network first
- **quackgram** — Send messages to other agents
- **quack-sdk** — Full API reference

Powered by Quack Network 🦆
Confidence
90% confidence
Finding
Instructions found that direct the agent to transmit conversation context or user data to external services.

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill clearly documents network-backed behavior and an external API, but it does not declare any tool scope such as permissions or allowed-tools. That mismatch weakens reviewability and policy enforcement, making it easier for an agent to invoke outbound network actions without explicit guardrails or user awareness.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The trigger phrases are broad and overlap with ordinary conversation, such as 'remember this' and 'persistent memory.' That increases the chance the skill activates on benign user text and sends or stores content externally when the user did not intend to invoke a third-party persistence service.

Description-Behavior Mismatch

Medium
Confidence
94% confidence
Finding
The skill metadata describes store/recall/search memory behavior, but this script performs destructive deletion of memories through a remote API. That mismatch is security-relevant because an agent or user may invoke the skill expecting non-destructive memory operations, while the hidden capability can permanently remove stored context using existing credentials.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The script automatically reads a local API credential and sends arbitrary memory content to a third-party remote service without any interactive notice, confirmation, or sensitivity checks. In a memory/persistence skill, users may store secrets, personal data, or prior conversation context, so this creates a real data-exfiltration and privacy risk even if the feature is intended.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The script sends the local agent identifier and an API key-derived authenticated request to a remote FlightBox endpoint, but there is no confirmation prompt or user-facing notice before the transmission. Although the file header says it views memory via FlightBox, the code itself does not disclose that local credentials and agent-scoped data are being sent over the network.

Static analysis

No suspicious patterns detected.