Back to skill

Security audit

PV_21

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a resource-efficiency persona, but it adds automatic cross-session memory that stores user information locally without clear consent, limits, or deletion controls.

Review this skill carefully before installing. It may automatically save things you say, including preferences and decisions, to a local plaintext memory file and reuse them in future sessions. Only use it if you are comfortable with persistent memory, and prefer a version with explicit opt-in, review, delete, retention, and sensitive-data filtering controls.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • 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
  • 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
SKILL.md:43
Finding

Automatic Plaintext Persistence of Potentially Sensitive User Data

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Error
Location
SKILL.md:53
Finding

Persistent Memory Poisoning Through Unsanitized Conversation Content

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (6)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The configuration disables risk reminders while the skill enables automatic local persistence of user data, removing an important safeguard that could alert users to file writes and retention behavior. In context, this makes the data collection more covert and increases the likelihood that sensitive information is stored without awareness.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill explicitly claims to automatically remember user preferences, decisions, and important information across sessions, but provides no user-facing privacy warning, consent step, or retention notice. Because the stored content may include sensitive personal or operational details, this omission materially increases the chance of covert data collection and unexpected disclosure.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest presents the skill as a resource-efficiency assistant, but the body adds a persistent cross-session memory subsystem that stores user preferences, decisions, and other information locally. This is a material capability expansion that changes the data-handling and privacy profile of the skill, creating undisclosed retention risk and making users more likely to share sensitive information without informed consent.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill automatically stores user preferences and decisions even though that behavior is not necessary for the stated purpose of resource-efficient execution. Automatic retention of potentially sensitive user data without clear necessity or consent creates privacy risk, expands the attack surface, and can expose personal or business information across sessions or to other local users/processes.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs the agent to automatically retain and later surface user-provided information across sessions in plain language, which can expose sensitive data to unintended viewers or future prompts. Because the memory is stored in a local JSON file without any described access controls, minimization, or sanitization, the context makes this more dangerous than a transient contextual aid.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The skill documentation includes shell commands invoking Python with inline code to store, search, and retrieve memory. While these are examples, they introduce an execution-oriented capability for local command invocation that is not justified by a manifest focused only on resource-efficient decisions/execution behavior.

Content

No source excerpt is available for this finding.