Back to skill

Security audit

Pansxng Websearch

Security checks for vulnerabilities and agentic risk

Overview

This skill provides local meta-search, but ordinary search requests can automatically install software, modify shell configuration, and start local services.

Install only if you are comfortable with a search request changing the host system: it may install Homebrew, Python, Valkey, and SearXNG, modify ~/.zshrc, pull unpinned code from the network, and leave local services running. Prefer a version that separates setup from search, asks for explicit confirmation, pins and verifies dependencies, uses a private app directory, and provides a clear uninstall path.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
Findings (4)

T03 · Remote Payload Retrieval and Execution

Error
Location
scripts/search.py:46
Finding

Mutable Remote Homebrew Installer Downloaded and Executed Without Verification

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Error
Location
scripts/search.py:218
Finding

Unpinned SearXNG Source and Python Dependencies Are Installed and Executed

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/search.py:265
Finding

Publicly Known Static Secret Assigned to Every SearXNG Installation

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/search.py:310
Finding

Predictable Shared Temporary Log Path Permits Symbolic-Link Redirection

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (17)

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The trigger phrases are broad and overlap with ordinary language such as '搜索', '帮我搜', and 'xxx是什么', making accidental activation likely. In this skill's context, accidental activation is more dangerous than usual because invoking the skill may lead to package installation, service startup, and other system changes rather than a harmless read-only lookup.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The code downloads and executes a remote Homebrew installer script via curl and bash without any confirmation. This is dangerous because it grants code execution based on network content and trust in an external repository, making compromise of the upstream source, transport, or environment highly impactful.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

Automatically appending to ~/.zshrc persists changes to the user's shell environment without consent. Persistent profile modification can alter future sessions, create hard-to-diagnose side effects, and is especially risky when done by a skill whose advertised function is merely searching the web.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill documentation describes capabilities that imply shell execution and filesystem modification, but it declares no explicit tool scope or permission boundaries. For a skill that can install packages, clone software, and manage services, the absence of scoped permissions materially increases the chance of unintended or excessive system access.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Although presented as a web search skill, the documentation states it will automatically install Homebrew, Python, Valkey, and SearXNG, which are significant system-modifying actions unrelated to a simple search request. This expands the trust boundary from 'search the web' to 'change the host environment,' creating risk of unauthorized software installation, persistence, and supply-chain exposure.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill states that first use will automatically install multiple system components, yet it does not provide an explicit warning or consent checkpoint for these host-modifying actions. Users asking for a search may unknowingly authorize package manager installation, background services, and software downloads, which is a serious mismatch between user intent and system impact.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The file description and user-facing output are Chinese-centric, and the CLI default language is set to zh-CN. This may violate a language/locale policy when no user opt-in or explicit locale selection is provided as the default interaction mode.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/search.py (reported line 26)May include surrounding context.

python
# ═══════════════════════════════════════════════════════════════

def _run(cmd, timeout=120, **kw):
    return subprocess.run(cmd, capture_output=True, text=True, timeout=timeout, **kw)


def _popen(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw):

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/search.py (reported line 30)May include surrounding context.

python
def _popen(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, **kw):
    return subprocess.Popen(cmd, stdout=stdout, stderr=stderr, **kw)


# ═══════════════════════════════════════════════════════════════

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

A search skill should not silently install a package manager and alter shell startup files during normal use. This expands the skill's privileges far beyond user expectations and executes externally fetched code, creating a large supply-chain and integrity risk if the remote installer or network path is compromised.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Installing and daemonizing Valkey in the background is an unjustified capability escalation for a web-search skill. It creates a persistent local service and increases attack surface without informed user consent, especially since the code proceeds automatically as part of making search work.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The stop function uses pkill -f searx.webapp, which can terminate processes system-wide based on a pattern match rather than a tracked PID. In a search-focused skill, that is a non-obvious and potentially disruptive capability that could kill unrelated processes whose command line matches the pattern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The search function sets lang='zh-CN' by default and always includes that locale in outgoing search parameters. This imposes a specific language/locale choice on users unless they override it manually, which matches the policy's language constraint concern.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

A normal search request can trigger installation of Homebrew, Python, Valkey, git clone operations, dependency installation, and service startup. Hidden side effects on a routine search path violate least surprise and least privilege, and they increase the likelihood that users authorize or trigger risky system changes unintentionally.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The option description specifies a default language of “zh-CN”, which sets a locale preference without indicating user choice or opt-in. This may conflict with organizational language/locale policy when the skill is used by users expecting locale neutrality or explicit selection.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The skill advertises itself as zero-configuration search, but also exposes install/start/stop/status lifecycle controls for local services. This is dangerous because users or calling agents may invoke what seems like a benign search skill while actually gaining operational control over background services and installation flows.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill clones third-party code into the user's Downloads directory and rewrites application configuration files while presenting itself as zero-configuration search. That mismatch in disclosure matters because it obscures material side effects and introduces supply-chain risk from pulling and modifying a large external codebase at runtime.

Content

No source excerpt is available for this finding.