boss直聘自动化（无GUI）

• Outcome: Review
• Latest audit: May 29, 2026 · 9:36 PM UTC
• Version: 1.0.0

Overview

This appears to be a real Boss Zhipin automation skill, but it can repeatedly control your desktop and send recruiter messages with exaggerated claims without review.

Install only if you are comfortable with a tool controlling your live desktop and sending job-platform messages on your behalf. Remove exaggerated claims from the template, use a test account or isolated browser window first, and add manual review/approval before any recruiter message is sent.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands

Findings (19)

Context-Inappropriate Capability

Medium

• Confidence: 96% confidence
• Finding: The skill explicitly instructs generated outreach to exaggerate qualifications ('了解→熟悉，不熟悉→了解'), which promotes deceptive impersonation in job applications. In the context of an automation tool that can automatically send messages at scale, this materially increases the risk of fraud, misrepresentation, and reputational or account consequences for the user.

Intent-Code Divergence

Medium

• Confidence: 91% confidence
• Finding: The documentation presents the system as doing 'intelligent matching' while later instructing the message generator to exaggerate applicant qualifications, creating a misleading trust boundary between analysis and communication. This can cause users or recipients to rely on outputs that appear objective but are intentionally embellished, making the automation more deceptive in practice.

Intent-Code Divergence

Medium

• Confidence: 95% confidence
• Finding: The comment understates the operational impact of the flow: it does not merely proceed without confirmation, it actively sends a message to an employer. In an automation skill that interacts with third-party services, misleading documentation can cause users to trigger unintended outbound communications, creating privacy, reputational, and policy-compliance risks.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The document explicitly describes automated communication actions including clicking a chat button, pasting generated text, and sending messages, but provides no warning, consent requirement, or rate/approval control. This creates a real risk of non-consensual bulk outreach, account abuse, and policy violations on third-party platforms.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill documentation states it captures screenshots and performs OCR on on-screen job content, but omits any notice about privacy, data handling, or the possibility of capturing sensitive information visible on screen. In an automation context, this can expose personal, confidential, or third-party data without informed user awareness.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The README explicitly advertises automatic message sending to employers, but does not disclose consent, review, rate-limiting, or platform-policy safeguards. In a job-application context, autonomous outreach can impersonate the user, spam recruiters, and cause unintended disclosure or reputational harm.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The README describes UI automation that scrolls listings, clicks jobs, takes screenshots, and performs OCR on captured content, yet provides no warnings about misclicks, sensitive on-screen data capture, or operating on the wrong window. Because this is desktop/UI automation, mistakes can affect unrelated applications and expose private information beyond the intended job page.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The skill documents desktop automation, screenshots, OCR, and automatic message sending without warning users about privacy, consent, or unintended actions on their system. Because pyautogui-style automation operates on the live desktop, misuse or mis-clicks can capture sensitive information or send messages without meaningful user awareness.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The quick-start commands enable immediate automated communication and browser control, but they do not present any guardrails, confirmation steps, or warnings about external side effects. This lowers the barrier to accidental misuse, including unsolicited outreach, unintended clicks, and privacy-impacting screenshots.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The paste_and_send function can transmit clipboard or supplied text directly to a live chat with no confirmation, preview, recipient validation, or safety interlock. In a GUI automation context, misfocused windows, poisoned clipboard contents, or unintended invocation could cause accidental data leakage, spam, or sending sensitive information to the wrong party.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The 'chat' command performs an outbound action that can send a message immediately after UI interaction, without an explicit confirmation prompt, dry-run mode, or warning to the operator. In an automation skill that interacts with a live job platform, this increases the chance of accidental spam, unintended communications, or policy-violating behavior if invoked mistakenly or scripted by another tool.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The code sends recruiter messages automatically with no user confirmation, review step, rate limiting, or recipient validation. In this context, the skill is designed to contact employers on a job platform, so unintended or repeated outreach could spam third parties, leak personal information from templates, violate platform rules, and damage the user's account or reputation.

Ssd 2

Medium

• Confidence: 98% confidence
• Finding: The documentation instructs the system to generate application messages that 'appropriately exaggerate' qualifications and imitate human language habits while expressing strong hiring interest. In context, this is deceptive automation aimed at misrepresenting the sender to recruiters, which materially increases social-engineering risk and makes the skill more dangerous than ordinary messaging automation.

Ssd 4

Medium

• Confidence: 97% confidence
• Finding: The README instructs the system to exaggerate qualifications in generated job outreach, normalizing deceptive self-presentation during employer communication. This is dangerous because it automates misrepresentation at scale, increasing fraud, reputational damage, and potential account or employment consequences for the user.

Ssd 4

Medium

• Confidence: 97% confidence
• Finding: The instructions normalize deceptive self-presentation by telling the system to overstate the applicant's abilities in generated messages. In a hiring context, especially when paired with automatic sending, this facilitates scaled deception toward third parties and can lead to fraud-like behavior, platform abuse, and reputational harm.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Boss 直聘职位沟通自动化依赖

# Python 依赖
pyautogui>=0.9.5
pyperclip>=1.8.2
Pillow>=10.0.0

• Confidence: 93% confidence
• Finding: pyautogui>=0.9.5

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Python 依赖
pyautogui>=0.9.5
pyperclip>=1.8.2
Pillow>=10.0.0

# 可选依赖

• Confidence: 92% confidence
• Finding: pyperclip>=1.8.2

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Python 依赖
pyautogui>=0.9.5
pyperclip>=1.8.2
Pillow>=10.0.0

# 可选依赖
pytesseract  # 如果使用 pytesseract OCR

• Confidence: 97% confidence
• Finding: Pillow>=10.0.0

Known Vulnerable Dependency: Pillow — 10 advisory(ies): CVE-2016-2533 (Pillow buffer overflow in ImagingPcdDecode); CVE-2023-50447 (Arbitrary Code Execution in Pillow); CVE-2021-27922 (Pillow Uncontrolled Resource Consumption) +7 more

Critical

• Category: Supply Chain
• Confidence: 98% confidence
• Finding: Pillow

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.