Back to skill

Security audit

PLUR Memory

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent persistent-memory tool, but it should be reviewed carefully because it automatically stores and reinjects memories across sessions and tools with limited scoping or approval controls.

Install only if you deliberately want a shared, persistent memory layer for your agents. Review the @plur-ai/claw package source and version before installation, avoid storing secrets or sensitive operational details as memories, and make sure you understand how to inspect, delete, disable sync, and limit shared ~/.plur/ memories.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:9
Finding

Unpinned Third-Party Plugin Installation

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Error
Location
SKILL.md:20
Finding

Uncontrolled Persistent Memory Propagation Across Sessions and Tools

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Memory Manipulation

High
Category
Memory Poisoning
Confidence
89% confidence
Finding

The skill instructs the agent to automatically learn from user corrections and inject relevant engrams every session, which creates a memory-poisoning pathway: malicious or mistaken inputs can be persisted and later reintroduced as trusted context. In a cross-session, cross-tool memory system, poisoned memories can influence future decisions well beyond the original conversation, increasing the blast radius of bad data or adversarial instruction shaping.

Content

Scanner excerpt · SKILL.md (reported line 24)May include surrounding context.

md
## Memory Lifecycle

- **Automatic injection** runs every session start - relevant engrams appear in your context
- When you discover something worth remembering - call `plur_learn` with a clear statement
- When corrected by the user - call `plur_learn` immediately with the correction
- When an injected engram was helpful - call `plur_feedback` with signal "positive"
- When an injected engram was wrong or stale - call `plur_feedback` with signal "negative"

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill explicitly describes storing corrections, preferences, and patterns as local YAML 'engrams' and later states that this memory is shared across tools and can sync across devices, but it does not present a prominent warning about the persistence and propagation of user-derived data. This creates a real privacy and data-governance risk because users or downstream agents may cause sensitive preferences, corrections, or operational details to be retained and redistributed without informed consent.

Content

No source excerpt is available for this finding.