Back to skill

Security audit

boss-geek

Security checks for vulnerabilities and agentic risk

Overview

This skill appears purpose-built for BOSS Zhipin job-search automation, but it asks users to install mutable third-party tooling that can use their logged-in browser session to read chats and send messages.

Review this before installing. Use a dedicated Chrome profile logged in only to BOSS Zhipin, inspect or pin the OpenCLI package and plugin revision, and manually confirm recipients and message text before running greet or send commands. Avoid using it from a browser profile that contains unrelated accounts or sensitive sessions.

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

Error
Location
SKILL.md:12
Finding

Unpinned Remote Dependencies Receive Access to an Authenticated Browser Session

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 12–22
Vulnerability Type: Unpinned and unverified third-party dependencies
Risk Level: High

Vulnerable Code Snippet

bash
npm install -g @jackwener/opencli
text
https://github.com/jackwener/opencli
bash
opencli plugin install github:SPYQWER1/opencli-plugin-boss-geek

The surrounding instructions require the user to log in to zhipin.com through Chrome and install the OpenCLI Chrome extension before installing the remote plugin.

Technical Analysis

The Skill directs users to install OpenCLI globally without specifying an exact package version. It also installs a plugin directly from a mutable GitHub repository reference without pinning a reviewed commit hash, release tag, checksum, or cryptographic signature.

Consequently, the code installed when a user follows these instructions may differ from the code that existed when the Skill was audited. The plugin is intended to reuse an authenticated Chrome session and can perform security-sensitive operations such as reading conversations, greeting recruiters, and sending messages. This combination creates a supply-chain trust boundary: a compromised package registry account, GitHub account, repository, release process, or upstream dependency could introduce arbitrary code with access to both the local execution environment and the authenticated browser session.

The audit did not establish that either upstream project is currently malicious. The confirmed issue is the absence of dependency pinning and integrity verification despite the dependency's sensitive access.

Attack Path

  1. An attacker compromises the npm package, GitHub repository, maintainer account, release pipeline, or one of the upstream dependencies.
  2. The attacker publishes malicious code under the same package or repository reference.
  3. A user follows the Skill instructions and runs the unpinned global npm installation or mutable GitHub plugi ...[truncated 1253 chars]
Remediation
View remediation

Remediation Suggestions

  1. Pin @jackwener/opencli to a specific reviewed version instead of installing the latest mutable release.
  2. Pin the GitHub plugin to a reviewed immutable commit hash rather than the repository's default branch.
  3. Publish and verify cryptographic checksums or signatures for all installed artifacts.
  4. Review the OpenCLI and plugin source code at the pinned revisions before granting browser access.
  5. Document the exact permissions and browser data available to the extension and plugin.
  6. Use a dedicated Chrome profile containing only the BOSS Zhipin session, with unrelated accounts and extensions removed.
  7. Avoid global installation where possible. Run the tool in a restricted project environment, container, or sandbox with minimal filesystem and network access.
  8. Require explicit user confirmation before operations that send greetings or chat messages.
  9. Establish a controlled upgrade process in which new dependency versions and commits are reviewed and integrity metadata is updated before deployment.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly supports actions like greeting HRs and sending chat messages using the user's existing logged-in BOSS直聘 session, but it does not clearly warn that these commands will perform external account actions on the user's behalf. This can lead to unintended outreach to third parties, reputational harm, or accidental misuse if a user invokes the skill without understanding that it is not just reading data but actively contacting others.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documented chatlist and chatmsg commands access conversation lists and message history from the user's authenticated BOSS直聘 account, which may contain sensitive personal and employment-related information. Without a privacy warning or data-handling guidance, users may expose private communications to the agent or downstream tooling without informed consent.

Content

No source excerpt is available for this finding.