Back to skill

Security audit

瑜伽学员管理

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent for yoga student management, but it stores email credentials locally in plaintext and can email sensitive student records without strong safeguards.

Review this skill before installing if you will store real student information. Use a dedicated app password, keep the skill directory private, avoid putting config.json or yoga_students.db in source control or backups, and be careful with the full-report email because it may send all student contact details and notes to the configured mailbox.

Vulnerability Patterns
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
yoga_student_manager/config.py:14
Finding

SMTP credentials are stored in plaintext with no enforced access restrictions

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
yoga_student_manager/database.py:207
Finding

Course-hour APIs accept negative values and permit balance and audit-log corruption

Content
View full analysis
Remediation
View remediation
= 0` - `remaining_hours IS NULL OR remaining_hours >= 0` - `deducted_hours > 0` - For hourly courses, require valid non-null totals and balances. 4. Consider enforcing `remaining_hours <= total_hours` if that invariant matches the business model. 5. Use a schema migration to rebuild existing SQLite tables with the new constraints. 6. Before migration, scan for negative deductions, negative balances, and inconsistent totals and repair them with an auditable process. 7. Add unit tests covering zero, negative, Boolean, floating-point, string, oversized, and insufficient-balance inputs. 8. Keep the balance update and deduction-log insertion in one transaction, and consider a conditional atomic update to reduce concurrent lost-update risks. ]]>
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill advertises student-management and deduction emails, but it also exposes a full-report email capability that can transmit all student records externally while lacking any declared permissions/tool scope for network and config access. In a system handling personal data, undeclared outbound data transfer materially increases the risk of unnoticed exfiltration or operator surprise.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

The skill advertises student-management and deduction emails, but it also exposes a full-report email capability that can transmit all student records externally while lacking any declared permissions/tool scope for network and config access. In a system handling personal data, undeclared outbound data transfer materially increases the risk of unnoticed exfiltration or operator surprise.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill describes capabilities that imply local file/database reads and writes, but it does not declare any tool scope such as permissions or allowed-tools. This weakens reviewability and sandbox enforcement because operators cannot clearly see or constrain the skill's storage access before use.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill stores student personal information and email configuration locally, including contact details and likely SMTP credentials, but the documentation does not clearly warn users that this sensitive data is persisted on disk. Without that notice, users may deploy it on shared or insecure machines and expose private records or credentials.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill automatically sends class-hour records by email, which is a network transmission of student-related information, but it does not provide a clear privacy notice or explain what data leaves the device. Automatic outbound transmission is more sensitive here because the context involves personal student records and recurring event logs.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill explicitly manages and stores student personal data (name, phone, WeChat, notes) and emails activity records, but the documentation provides no privacy warning, data-handling notice, or guidance on protecting local database and email contents. This increases the risk of inadvertent disclosure, insecure operator behavior, and noncompliance with privacy expectations when handling personally identifiable information.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The code stores SMTP credentials, including the email password or app authorization code, directly in a local JSON file in plaintext. If the local machine, project directory, backups, or source distribution are accessed by another user or malware, these credentials can be recovered and used to access the email account or send fraudulent messages.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The natural-language strings for configuration prompts and status messages are entirely in Chinese, and the file does not provide any user opt-in or alternate locale. Under the policy, forcing a specific language without choice is a natural-language policy violation unless the locale restriction is explicitly documented and justified.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The delete method removes a student and explicitly deletes related deduction logs and course records before committing the transaction. In this file there is no confirmation prompt, logging/print disclosure, or explanatory docstring/comment warning that the operation is destructive and cascades to associated data.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The email body is assembled with direct inclusion of sensitive student contact details and free-form notes, which may contain additional confidential or special-category information. In this skill context, a yoga teacher management system handles real student records, so transmitting these details in ordinary email materially raises privacy and confidentiality risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

This function emails student records containing personal data, including names, phone numbers, WeChat IDs, notes, course details, and deduction logs, without any visible consent check, minimization, or warning at the point of transmission. Even if SMTP uses TLS in transit, sending broad PII by email increases exposure through mailbox compromise, misdelivery, forwarding, and retention in plain-text email stores.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This code hard-codes all user-facing prompts and labels in Chinese, which imposes a specific language on all users. The file does not offer a language selection mechanism or explain that the tool is intentionally limited to a Chinese-speaking context.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Nearly all user-visible strings in this file are fixed Chinese messages, and there is no indication of locale selection or user opt-in. This creates a natural-language policy issue because the skill appears to enforce a single language for all users.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Deleting a student record without confirmation or warning can cause irreversible loss of student and course-management data through operator error or misuse. In this application, deletion may also remove historical records needed for billing, attendance tracking, or dispute resolution, making accidental destruction materially harmful.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The code provides a separate send_full_report() path that emails all student data, which goes beyond the stated behavior of sending a record after hour deduction. This expands the data-exfiltration surface for personally identifiable student information and increases the blast radius if the feature is misused, triggered accidentally, or configured with the wrong recipient.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Bulk emailing of all student records is not necessary for routine lesson-hour management and creates an unnecessary channel for transmitting sensitive data. In this skill context, the dataset likely includes names, phone numbers, WeChat IDs, notes, and course history, so a single action can disclose the full local database externally.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

文件中的名称与描述均将技能定位为中文使用场景,且未说明是否支持其他语言或允许用户选择语言/地区设置。根据语言/locale 政策,若技能默认强制特定语言而无用户选择,属于自然语言层面的潜在政策问题。

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The documented menu includes student deletion, but there is no warning that deletion may be irreversible and no guidance to require confirmation or backup before destructive actions. In a system storing business and student records, this can lead to accidental loss of data and audit history rather than direct compromise.

Content

No source excerpt is available for this finding.