Back to skill

Security audit

摇摇记忆系统

Security checks for vulnerabilities and agentic risk

Overview

This memory skill is coherent in purpose, but it enables persistent storage and IMA synchronization in ways that can transmit private memory data without reliable user control.

Review this carefully before installing. Use it only if you are comfortable with conversation-derived memories being stored on disk and potentially synchronized to IMA. Do not rely on enabled=false as a sync block in the current script, avoid using --file with sensitive paths, and keep secrets, credentials, private keys, and confidential project files out of memory content.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (3)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/sync_ima.py:120
Finding

Unrestricted File Path Allows Upload of Arbitrary Readable Files

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/sync_ima.py:42
Finding

Documented Synchronization Disable Flag Is Not Enforced

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
SKILL.md:235
Finding

Installation Instructions Execute an Unpinned Registry Package

Content
View full analysis
Remediation
View remediation
install yaoyaoya-memory ``` 2. Where supported, verify the downloaded package with an integrity digest or signed release. 3. Document the expected package registry and publisher identity. 4. Use a lockfile or a controlled installation environment for repeatable deployments. 5. Review package lifecycle scripts before execution. 6. Update the pinned version only after reviewing and testing the new release. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (14)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README advertises persistent memory and says IMA knowledge-base sync is enabled by default, but it does not clearly warn users that data may be stored long-term and transmitted to an external platform. Because the skill is specifically designed to capture cross-session memories, the lack of prominent consent and disclosure materially increases the chance of privacy violations or unintended exfiltration of sensitive information.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

Allowing the AI to decide on its own that information is 'worth long-term saving' delegates consent and retention decisions to an unreliable heuristic. In this skill's context, that can lead to persistent storage of secrets, personal data, or confidential project details even when the user never explicitly asked for retention.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The IMA sync feature is enabled by default and synchronizes user preferences, decisions, and project data to an external platform, yet the skill does not prominently warn that user data may leave the local environment. This significantly increases privacy and confidentiality risk because the skill handles memory content that may include personal or sensitive information.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The retrieval triggers like '我记得、之前说过' are ambiguous and can overlap with normal dialogue, causing the assistant to access persistent memory when the user may only be speaking conversationally. In this skill context, that can surface stale or sensitive stored data unexpectedly, creating privacy and context-integrity risks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The memory-write trigger phrases such as '记住、记录、存档' are broad enough to match ordinary conversation, which can cause unintentional persistence of sensitive user content. In a memory skill, accidental writes are particularly risky because they create durable storage and may later be synchronized externally.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The README instructs users to run npx clawhub install yaoyaoya-memory without pinning a specific version. This allows execution of whatever package version is current at install time, increasing supply-chain risk if the package or one of its transitive dependencies is compromised or unexpectedly changed.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

Most of the user-facing instructions, trigger phrases, labels, and examples are presented only in Chinese, which effectively constrains usage to a specific language. The file does not offer multilingual alternatives or state that the locale restriction is intentional and justified.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill describes writing conversation-derived information into MEMORY.md, daily logs, and learning files, but does not present a clear user-facing warning that this storage is persistent. Users may disclose information under normal chat expectations without realizing it will be retained across sessions on disk.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The memory-write trigger phrases are broad enough to overlap with ordinary conversation, so the assistant may persist data when the user did not intend durable storage. In a memory skill, accidental writes can capture sensitive personal, project, or behavioral information into local files without meaningful consent.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill's default behavior semantically encourages persistence and synchronization of user profile data, preferences, decisions, and project details. In context, this is more dangerous because the skill is specifically designed as a long-term memory layer, making privacy leakage and over-retention a core operational risk rather than an incidental one.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
80% confidence
Finding

The install command references an unpinned package via npx clawhub install yaoyaoya-memory, which can fetch whatever version is current at execution time. This creates a supply-chain risk where a compromised or incompatible upstream release could be installed unexpectedly.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The notes_mapping keys are natural-language labels written only in Chinese, which suggests the skill is configured around a fixed language/locale. There is no indication in this file that users can choose another language or that the Chinese-only constraint is intentionally limited to a region-specific deployment.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This code file contains natural-language instructions and messages that enforce a specific language/locale for users. The policy allows locale constraints only when the user is given a choice or the restriction is clearly justified, which is not present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This code file contains user-facing natural-language instructions and examples in Chinese, and the CLI also includes Chinese note examples and prompts. Because the skill does not offer a language choice or document a justified locale restriction, it may violate a language/locale policy requiring user opt-in.

Content

No source excerpt is available for this finding.