Back to skill

Security audit

Taplio LinkedIn Connection Request Hook

Security checks for vulnerabilities and agentic risk

Overview

The scanners report a benign verdict with two low-probability findings related to outreach behavior, while one referenced item could not be resolved and the evidence is incomplete.

Consider whether you are comfortable with a note-writing skill also being able to queue invitations and choose the target's language over your default. The verdict is benign, but the scan was partial: a local path-like reference could not be resolved, so the available evidence does not cover everything.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

The skill can queue LinkedIn invitations despite being described as a note-writing skill.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest at L003 describes generating three connection-request notes and retrieving context through Taplio. L098 additionally authorizes calling send_connection_request(username) to queue an actual invitation without a note, introducing a remote account action beyond drafting. Requiring explicit approval limits unintended execution but does not resolve the mismatch with the declared scope.

Content

No source excerpt is available for this finding.

Requires the target's language even when it differs from the user's default, without user opt-in.

Low
Category
Output Handling
Confidence
96% confidence
Finding

The rule explicitly overrides the user's default language with the target's language. Although this may help tailor outreach, the skill does not offer a language choice or request consent to the override.

Content

No source excerpt is available for this finding.