Back to skill

Security audit

perplexity-research-zh

Security checks for vulnerabilities and agentic risk

Overview

The skill is not plainly malicious, but its bundled client can perform broader AISA web, Tavily, URL extraction, and multi-source search workflows than the Perplexity-only skill description discloses.

Review before installing. This skill should only be used if you are comfortable giving it an AISA API key and sending research queries or supplied URLs to AISA-backed services. The main concern is disclosure: the package advertises Perplexity Sonar research, but the included client has additional search and extraction commands that should be documented, removed, or split into separate skills.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The documented purpose is narrowly framed as Perplexity Sonar deep research, but the underlying client reportedly supports broader search modes, URL content extraction, and multi-source aggregation. That mismatch is security-relevant because operators may grant trust and permissions based on the declared narrow function while the actual implementation can retrieve arbitrary remote content or perform broader network operations than expected.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file header and documented usage show a CLI that supports multiple search engines, URL extraction, and non-Perplexity workflows, which materially exceeds the manifest’s stated purpose of Perplexity Sonar deep-research answers. In an agent-skill setting, this scope mismatch is dangerous because it can enable undisclosed outbound data flows and capabilities that callers, reviewers, or policy controls did not authorize.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

These command handlers implement web, scholar, smart, Tavily, and extract operations that are not reflected in the skill description. That hidden functionality broadens the attack surface and could transmit user queries or supplied URLs/content to services outside the user’s expected Perplexity-only research flow.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
85% confidence
Finding

The skill declares environment and network capabilities via required API key and an external search client, but it does not declare any explicit tool scope such as permissions or allowed-tools. This creates an authorization blind spot: callers and reviewers cannot easily tell what the skill is permitted to access, increasing the chance of unintended network use or secret exposure through loosely governed execution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill name and all user-facing documentation are presented only in Chinese, with no indication that language selection is optional or that the skill is region-specific. This can violate language/locale policy requirements because it effectively constrains usage to a specific language without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest sets the skill language to "zh" and the slug also encodes a Chinese variant, but this file provides no indication that the user can opt into another language or that the locale restriction is required for a region-specific purpose. This is a natural-language policy concern because it appears to enforce a specific language without documented choice or justification.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · scripts/search_client.py (reported line 28)May include surrounding context.

python
from concurrent.futures import ThreadPoolExecutor, as_completed
from typing import Any

AISA_BASE = "https://api.aisa.one/apis/v1"


def get_api_key() -> str:

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The verity workflow performs parallel retrieval across several sources, computes a confidence score, and sends aggregated results for AI synthesis, none of which is disclosed by the manifest. While functionally useful, this hidden orchestration increases external data sharing and can mislead users about the provenance, processing, and scope of their data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The code hard-codes the explain request language to "en", which enforces a specific language regardless of user preference. This is a natural-language policy concern because the file provides no user opt-in, locale selection, or justification for restricting synthesis output to English.

Content

No source excerpt is available for this finding.