Back to skill

Security audit

书搭子

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a local reading companion, but its optional voice feature permits external command/script execution while the manifest says external scripts and networking are disabled.

Review carefully before installing. The local reading and emotion journal behavior is disclosed, but do not enable voice output unless you trust the exact TTS command, understand that a custom script can run local code, and are comfortable with any network behavior of the chosen TTS tool. Treat the data directory as sensitive because it stores emotional notes, private code words, dates, and reading history in plaintext.

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 (2)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:49
Finding

Arbitrary Command Execution Through Configurable TTS Command

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
SKILL.md:84
Finding

Unpinned Third-Party Package Installation

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (8)

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README claims the skill is strictly self-contained and does not touch external paths, but it explicitly supports executing a TTS command from system PATH or an absolute external script path. That contradiction weakens the documented trust boundary and could mislead users into enabling execution of arbitrary external binaries or scripts under the skill context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The usage section describes broad, always-on behavior such as guiding profile creation, accepting emotional state input, auto-maintaining a library, and auto-logging each interaction, but it does not define clear trigger boundaries. In an agent skill, ambiguous invocation scope increases the risk of over-collection, unintended persistence, or the model acting on sensitive content when the user did not intend to invoke those workflows.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README says the AI automatically maintains files in the reading library and automatically records each interaction to emotion logs, but it does not present a clear upfront warning at the point of use about persistent storage side effects. Because the skill handles potentially sensitive reading habits and emotional data, silent persistence raises meaningful privacy and consent concerns even if storage is local.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The README states there are zero network requests, yet recommends edge-tts, which typically relies on external network services. This creates a false privacy/security assurance that may cause users to expose sensitive reading or emotional content to a remote provider without informed consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill explicitly stores user profile data including reading preferences, emotional support needs, private code words, and important dates in local markdown files, but it does not clearly warn users that this sensitive personal and mental-health-adjacent information will persist on disk. Even though storage is local-only, persistent plaintext storage can expose highly sensitive data to other local users, backups, indexing tools, or later compromise of the host environment.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs the agent to record emotion events and triggers into dated log files without a clear user-facing disclosure that mental-health-related content will be persistently stored. This creates privacy risk because emotional distress signals, crisis code words, and related notes are especially sensitive and may be retained indefinitely in accessible plaintext files.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest description is entirely in Chinese and presents the skill as a Chinese-language reading companion, but there is no indication that users can choose another language or that the locale restriction is required for a region-specific purpose. This creates a natural-language policy concern because it implies a fixed language without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The entire file is written in Chinese and presents the user profile schema only in that language, with no indication that the user can choose another language or locale. Under the stated policy, a fixed language choice without user opt-in is a natural-language policy issue unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.