Back to skill

Security audit

research-assistant

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Bear-note automation that reads tagged notes, searches for GIFs from note-derived keywords, appends media, and removes the processing tag.

Install only if you are comfortable letting the agent read Bear notes tagged "待整理", derive search keywords from them, query GIF/web tools, append GIF markdown, and remove that tag. Consider backing up important notes or asking for a preview before edits when notes are sensitive or business-critical.

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 Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description claims an end-to-end note-finalization workflow, but the supplied code chunk only fetches and enumerates tagged Bear notes and their titles. It explicitly states that keyword extraction, GIF search, GIF insertion, and tag removal are handled elsewhere by an agent. So the primary implemented behavior is note discovery/listing, not auto-illustration or finalization. Resource access is broadly consistent with Bear notes, but the key advertised capabilities are absent from this code.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill description does not clearly warn that it will automatically modify Bear notes and remove the 待整理 tag, which can cause unintended data changes or premature workflow state transitions. Users may invoke the skill without understanding that it performs write operations, increasing the risk of accidental note corruption, unwanted edits, or loss of review markers.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The workflow derives keywords from note content and may send them to external GIF search tools or web services, but it does not disclose this privacy implication. If notes contain sensitive research topics, internal project names, or personal data, those terms could be exposed to third parties through search queries.

Content

No source excerpt is available for this finding.