Back to skill

Security audit

WiseDiag-Checkup

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed cloud-based medical report analyzer, but users should understand that it uploads health documents to WiseDiag and saves analysis results locally.

Install only if you are comfortable sending checkup reports, questionnaire details, and optional member IDs to WiseDiag's cloud API. Avoid using it for highly sensitive documents unless you trust WiseDiag's handling and retention claims, and save generated Markdown reports only in private locations. Consider pinning dependencies before production use.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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 (1)

T08 · Insecure Dependencies

Note
Location
requirements.txt:1
Finding

Unbounded Third-Party Dependency Permits Unreviewed Future Releases

Content
View full analysis

Vulnerability Details

File Location: requirements.txt:1
Vulnerability Type: Unpinned dependency and software supply-chain exposure
Risk Level: Low

Complete Code Snippet:

text
requests>=2.28.0

Technical Analysis

The dependency specification establishes only a minimum version and permits installation of any future requests release. Consequently, installations are not reproducible and may consume a release that has not been reviewed or tested by the project.

This is not evidence that the current requests package is malicious. Exploitation requires a future allowed release, its distribution channel, or the dependency-resolution environment to become compromised. If that occurs, malicious package code could execute during installation or when scripts/wise_analyze.py imports requests.

This risk is particularly relevant because the application handles medical reports and reads WISEDIAG_API_KEY from the process environment.

Attack Path

  1. An attacker compromises an allowed future requests distribution, its publishing account, or another relevant supply-chain component.
  2. A user runs pip install -r requirements.txt without a lock file or hash verification.
  3. The package resolver selects the compromised release because it satisfies requests>=2.28.0.
  4. Attacker-controlled code executes during package installation or when wise_analyze.py imports the dependency.
  5. That code operates with the installing or invoking user's privileges and may access readable local files, process environment variables, medical-report inputs, and network resources.

Impact Assessment

Successful exploitation could obtain the privileges of the user installing or running the Skill. Within that account's access boundaries, malicious dependency code could read WISEDIAG_API_KEY, access submitted medical documents and other readable files, alter generated reports, or transmit data over th ...[truncated 201 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin requests and all transitive dependencies to reviewed versions through a lock file.
  2. Generate and enforce cryptographic hashes, for example with pip-compile --generate-hashes and pip install --require-hashes.
  3. Install packages only from an explicitly configured trusted package index.
  4. Use automated dependency monitoring, but require review and testing before updating pinned versions.
  5. Build and install dependencies in an isolated environment with minimal filesystem access and without exposing WISEDIAG_API_KEY during installation.
  6. Maintain a documented process for promptly updating pins when security advisories affect the selected versions.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (10)

Session Persistence

Medium
Category
Rogue Agent
Confidence
90% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 60)May include surrounding context.

Temporary (current terminal session)

export WISEDIAG_API_KEY=your_api_key_here

Permanent (add to ~/.zshrc or ~/.bashrc)

echo 'export WISEDIAG_API_KEY=your_api_key_here' >> ~/.zshrc source ~/.zshrc

text

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill performs sensitive actions including reading environment secrets, uploading local medical files to a remote service, and saving files locally, but it declares no explicit tool scope or permission boundaries. In an agent environment, this can lead to overbroad invocation and unintended access to local files or credentials, especially given the highly sensitive medical context and remote network transmission.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger guidance is broad and includes vague phrases like 'or similar requests,' which can cause the agent to invoke the skill in situations the user did not clearly authorize. Because the skill uploads medical documents to a third-party cloud service, accidental activation increases privacy and data-exfiltration risk beyond ordinary over-triggering.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill instructs the agent to produce a fixed-language introduction/example ('以下是您的体检报告分析结果:') without checking the user's language preference. This is not a direct security flaw by itself, but in a medical-report workflow it can degrade user comprehension and informed consent around sensitive health output, increasing the chance of misunderstanding important findings.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

This code performs external HTTP POST requests to a remote API and, in this skill's context, those requests include sensitive medical report metadata, questionnaire text, task identifiers, and potentially uploaded files. External transmission is expected for the service design, but it remains a real data-exposure risk because the skill handles health information and does not enforce explicit informed consent or data-minimization safeguards.

Content

Scanner excerpt · scripts/wise_analyze.py (reported line 85)May include surrounding context.

python
for attempt in range(1, max_retries + 1):
        try:
            print(f"[*] Requesting API... (attempt {attempt}/{max_retries})")
            resp = requests.post(
                endpoint, json=payload, headers=headers,
                timeout=REQUEST_TIMEOUT,
            )

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The report formatter generates headings, labels, and the medical disclaimer entirely in Chinese, forcing a specific language for user-facing output. There is no option for language selection and no documented justification that this skill is intentionally limited to Chinese-speaking users or a China-specific context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This skill uploads medical checkup PDFs and questionnaire data to an external third-party API, but the CLI flow does not present an explicit consent, privacy, retention, or data-sharing warning at the point of transmission. Because the data is highly sensitive health information, users may unknowingly disclose protected personal data to an external service, creating privacy, compliance, and trust risks even if transport uses HTTPS.

Content

No source excerpt is available for this finding.

Unpinned Dependencies

Low
Category
Supply Chain
Confidence
95% confidence
Finding

The dependency is specified as requests>=2.28.0, which allows installation of any newer version without guaranteeing a tested or secure release. This weakens reproducibility and can result in deploying a version with known vulnerabilities or behavior changes, especially risky in a skill that may fetch user-supplied PDF URLs over the network.

Content

Scanner excerpt · requirements.txt (reported line 1)May include surrounding context.

text
requests>=2.28.0

Unverifiable Dependency: requests has 16 known advisory(ies) (CVE-2014-1830 (Exposure of Sensitive Information to an Unauthorized Actor in Requests); CVE-2024-47081 (Requests vulnerable to .netrc credentials leak via malicious URLs); CVE-2024-35195 (Requests `Session` object does not verify requests after making first request wi) +13 more), but the manifest does not pin a version, so it is unknown whether the installed release is affected

Low
Category
Supply Chain
Confidence
90% confidence
Finding

Because requests is not pinned, there is no assurance that deployment will avoid versions affected by published advisories. In this skill's context, which may retrieve remote PDFs from user-provided URLs, a vulnerable requests version could expose credentials, mishandle redirects, or introduce other network-related security issues when processing untrusted input.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The save command writes AI-generated medical analysis results to disk as Markdown in the current or user-specified directory without warning that sensitive health data will persist locally. On shared machines, synced folders, backups, or lax-permission directories, this can expose private medical information beyond the user's expectation.

Content

No source excerpt is available for this finding.