Back to skill

Security audit

gety-local-search

Security checks for vulnerabilities and agentic risk

Overview

This local-search skill is not malicious, but it needs review because it can expose and change the scope of locally indexed files without clear consent guidance.

Install only if you expect the agent to search local indexed files. Before allowing connector add or remove commands, confirm the exact folder or connector ID and understand that this can change what private local content Gety indexes in future searches.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger description is broad enough to match generic requests like '帮我搜索' or 'search' that may not actually be about local files. That increases the risk of unintended invocation, causing the agent to search local indexed content when the user may have meant web search or another safer/non-local action, potentially exposing private file metadata or contents.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill is described as a local file search/retrieval capability, but its documentation also exposes connector management commands that can change what directories are indexed or remove existing indexed sources. This expands the skill from read-oriented search into state-changing local system configuration, increasing the chance an agent modifies the user's indexing scope without clear authorization or necessity.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The connector add/remove instructions modify the set of local directories being indexed, but the skill provides no warning that these are persistent, privacy-relevant changes. Without explicit caution and consent requirements, an agent could silently broaden surveillance of the filesystem or disable expected access by removing connectors.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Including gety connector add and gety connector remove in a search-focused skill gives the agent unjustified authority to alter which local directories are indexed. An unintended invocation or prompt-driven misuse could cause privacy-impacting expansion of indexed data or disruption by removing connectors, which is outside the user's likely expectation for a simple search skill.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

This is a markdown file, so SQP-2 applies to user-facing descriptions of actions that could affect user data or system state. The '移除连接器' section states that a connector can be deleted but does not include any caution about what removal means for existing indexing, access, or whether the action is reversible.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.