Security audit

Meta Agent Optimizer Free

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent logging/memory tool, but it encourages automatic persistent storage of conversation and error details and promotion into agent instruction files without enough user control or sanitization.

Install only if you are comfortable with the agent writing persistent learning logs and potentially updating future-agent instruction files. Keep .learnings local or ignored by Git unless entries are reviewed and sanitized, and require explicit confirmation before promoting anything into CLAUDE.md, AGENTS.md, or Copilot instruction files.

A.I.G

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)

T02 · Agent Memory Poisoning

Error

Location: SKILL.md:252
Finding: Untrusted Conversation Content Can Poison Persistent Agent Instructions
Content: View full analysis
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Warning

Location: SKILL.md:155
Finding: Automatic Diagnostic Logging Can Persist Secrets in Shared Repository History
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (7)

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The skill instructs the agent to create persistent local files immediately, but it does not prominently warn that these writes may remain on disk and may later be committed to a repository. Because the logged material can include user corrections, errors, and context, this creates a realistic risk of unintended long-term retention and disclosure.
Content: No source excerpt is available for this finding.

Ssd 3

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill explicitly asks for full context, attempted commands, inputs, and user feedback to be recorded in reusable markdown logs. Those fields commonly contain secrets, tokens, credentials, internal paths, error dumps, customer data, or sensitive business context, so storing them verbatim creates a significant leakage and compliance risk.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The trigger phrases are broad enough to match normal conversation, causing the skill to activate and log content when the user may not have intended any persistence action. In a skill that can write files and promote content into project memory, accidental activation increases the chance of storing sensitive prompts, corrections, or operational details without meaningful consent.
Content: No source excerpt is available for this finding.

Ssd 3

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: This instruction tells the agent to persist user-provided corrections and knowledge into reusable logs, and later sections encourage promotion into durable project memory files like CLAUDE.md or AGENTS.md. That creates a plain-language data exfiltration and retention path where sensitive conversation content, internal conventions, or proprietary details can spread beyond the original interaction.
Content: No source excerpt is available for this finding.

Ssd 3

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The guidance explicitly recommends a team-wide strategy of tracking learning logs in the repository, which broadens access to whatever user or operational data those logs contain. In combination with the skill's encouragement to capture detailed context and feedback, repository persistence can turn transient chat or debugging information into widely shared, durable exposure.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The file states '完全中文化表达，适配中文用户工作流', indicating a language-specific design. There is no corresponding opt-in, alternate language option, or clear documentation that the locale restriction is required for a region-specific purpose.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Low

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The frontmatter declares license: Proprietary at L004, but the later License section states the improved work's license is MIT and says it complies with MIT requirements at L496-L511. This is an active documentation contradiction about the skill's licensing/usage intent, not merely missing detail.
Content: No source excerpt is available for this finding.