Back to skill

Security audit

Sleep Story

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent sleep-story generator, but it asks for sensitive sleep and preference tracking plus persistent agent memory in ways users should review before installing.

Install only if you are comfortable with hypnotic-style sleep content and persistent personalization. Disable or avoid storing sleep metrics, do not allow the skill to write user preferences into TOOLS.md, keep any profile local and reviewable, and provide explicit opt-in plus a simple way to delete all stored history and preferences.

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
references/feedback-loop-system.md:76
Finding

Indefinite Plaintext Storage of Sleep and Behavioral Data

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Warning
Location
SKILL.md:389
Finding

User-Controlled Preferences May Be Written into Shared Agent Configuration

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
  • 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
Findings (65)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill advertises hypnotic suggestion, preference tracking, and cross-session memory without any clear user notice, consent mechanism, or explanation of risks and data use. In a mental-wellness and sleep context, undisclosed persuasive techniques plus persistent profiling are more sensitive than ordinary creative assistance and can undermine user autonomy and privacy.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The roadmap proposes cross-user learning, community sharing, A/B testing, and multimodal or physiological feedback collection, all of which substantially increase data sensitivity and secondary-use risk beyond standalone story generation. In a sleep/mental-wellness setting, these features are especially dangerous because they enable aggregation, experimentation, and potential exposure of highly personal behavioral or bodily data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The document explicitly describes persistent storage of user preference and sleep-related data, but provides no notice, consent flow, retention policy, access controls, or minimization guidance. Because sleep patterns and psychological preferences can be sensitive personal data, silently collecting and retaining them creates privacy and compliance risk if the skill is deployed as described.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This section proposes automated feedback collection, sleep-effectiveness tracking, and data analysis workflows without informing users that their responses and inferred sleep outcomes may be monitored and processed. In the context of a sleep-assistance skill using psychological techniques, this increases the sensitivity of the data and the harm from undisclosed profiling or over-collection.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file uses a single language throughout and does not indicate that contributors may use another language or that the repository is intentionally Chinese-only. Under the language/locale policy rule, forcing a specific language without user opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The integration document expands the skill from simple sleep-story generation into a broader behavioral profiling and optimization system with persistent feedback collection, preference modeling, and effect validation. That creates a material scope mismatch: users invoking a bedtime story skill would not reasonably expect ongoing profiling and experimentation infrastructure unless it is clearly disclosed and justified.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This section documents cross-session memory and tracking of sleep effectiveness metrics such as sleep time and improvement trends, which are sensitive wellness-related data. Persisting that information without a clearly necessary purpose, consent model, retention rule, or minimization boundary increases privacy risk and enables unexpected longitudinal profiling.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The markdown specifies collecting detailed feedback, preferences, and sleep-effectiveness information, but it does not include any privacy notice, lawful basis, storage limitation, security handling, or user-control language. In a sleep-assistance context, this omission is particularly risky because the data relates to personal habits and mental/wellness state, making misuse or overcollection more harmful.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

This section introduces A/B testing, long-term tracking, and cross-user learning, including anonymized learning, without any user warning or meaningful transparency about experimentation and secondary data use. Even if cross-user learning is said to be anonymous, the lack of disclosure and consent makes the practice unsafe and inconsistent with user expectations for a simple sleep-story tool.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document describes A/B testing, control comparisons, and validation workflows on users even though experimentation is not part of the manifest’s stated purpose. Running optimization experiments in a sleep/wellness context without explicit notice and consent can expose users to undisclosed manipulation and collection of comparative outcome data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This markdown file presents all user-facing instructions and labels in Chinese only, which effectively forces a specific language for users submitting issues. The policy allows locale constraints only when the skill offers language choice or clearly documents a justified region-specific limitation, neither of which is present here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
78% confidence
Finding

The file is entirely written as a Chinese-only operational specification and presents the skill behavior as fixed, with no indication that users may choose another language or that the locale restriction is required for a region-specific purpose. This can constitute a language policy violation when a skill effectively forces a specific language without opt-in.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document explicitly broadens the skill from generating soothing stories into an active hypnosis/induction system with direct suggestion, consciousness-blurring, and resistance-bypassing techniques. That creates a material scope expansion beyond the stated storytelling purpose, increasing the risk of manipulative output and safety issues for vulnerable users seeking relaxation rather than hypnosis.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The workflow extends the skill into collecting and maintaining sleep effectiveness data and user preference/history records, which is meaningfully broader than simple story generation. This creates data-handling and profiling behavior not transparently described in the skill purpose, so users may be subjected to retention and optimization logic they did not knowingly agree to.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

Tracking user sleep outcomes, preferred story types, and which techniques are most effective is not clearly necessary for the stated task of producing a sleep story on demand. Because the data concerns mental state and sleep behavior, unnecessary collection increases privacy risk and creates a more sensitive user profile than the feature set appears to require.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The document instructs operators to record user sleep time, preferences, and effectiveness data without any notice, consent flow, or privacy safeguards. In this context, the information may reveal health-adjacent or psychological patterns, so silent collection and storage materially increases the risk of privacy harm, misuse, or noncompliant handling of sensitive personal data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This markdown file uses Chinese throughout for headings, descriptions, and workflow instructions, which effectively forces a specific language on users and contributors. Under the policy, a fixed language is only acceptable if the skill offers a language choice or clearly documents a justified locale-specific scope, neither of which appears here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The entire skill description, usage examples, and interaction framing are presented only in Chinese, with no statement that the skill is Chinese-only by design and no option for users to choose another language. Under the policy, forcing a specific language without user opt-in is a natural-language locale violation unless the restriction is clearly justified.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The example trigger phrases are very broad everyday expressions such as asking for a story or saying one is tired, which can cause the skill to activate in situations beyond the user's clear intent. In a sleep-focused skill that uses relaxation and hypnotic-style prompting, ambiguous activation increases the risk of unwanted behavioral influence or inappropriate invocation during unrelated conversations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README describes collecting user preferences, sleep feedback, trends, and long-term tracking, but does not disclose retention limits, consent flow, storage protections, or deletion controls. Because this concerns sensitive wellness and behavioral data, the omission creates privacy and compliance risk if users are profiled or their sleep data is stored without informed consent.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 297)May include surrounding context.

md
## 🌟 Star History

[![Star History Chart](https://api.star-history.com/svg?repos=yourusername/sleep-story-skill&type=Date)](https://star-history.com/#yourusername/sleep-story-skill&Date)

---

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

L003 的描述整体以中文定义技能输出和使用方式,且文档未说明用户可选择其他语言,也未将中文限制为特定地区或合规场景要求。这会构成默认语言/locale 强制,缺少用户显式选择。

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

L035 的“有什么放松的方法吗”和 L037 的“任何需要心理安抚、放松助眠的场景”范围过宽,覆盖大量普通对话请求,未清楚限定何时应激活本技能而非其他一般建议类技能。该描述缺少明确边界或排除条件,容易造成意外调用。

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill explicitly introduces cross-session personalization, preference tracking, feedback collection, and persistent memory for a bedtime-story generator. That exceeds the minimum data needed to produce a single story and creates unnecessary retention of behavioral and preference data without any clear consent, minimization, or retention limits.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill plainly instructs persistent collection and retention of user-specific preferences and history across sessions. This is dangerous because it normalizes broad behavioral data retention for a low-necessity use case and can expose personal routines, emotional states, and content preferences if storage is mishandled.

Content

No source excerpt is available for this finding.