ClawHub
Back to skill

Security audit

Taplio LinkedIn Post Performance Critic

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a coherent LinkedIn draft review using disclosed Taplio account data, topic searches, and optional draft saving.

Review the Taplio connector permissions before connecting. Benchmark searches disclose your draft's topic to Taplio, so generalize confidential details. Review proposed changes before accepting a save; publishing requires your explicit confirmation.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (3)

The benchmarking step sends the draft's topic to Taplio without a privacy disclosure.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The mandatory benchmarking step transmits a topic derived from an unpublished draft to an external service without a sensitivity check or query-minimization requirement. A topic can reveal confidential product plans, clients, or internal events even without sending the full draft; the explicitly advertised Taplio integration makes ordinary benchmarking expected, but does not adequately address sensitive draft-derived queries.

Content

No source excerpt is available for this finding.

The skill adds remote draft creation and modification to a manifest-described critique workflow.

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The manifest describes auditing a draft and returning scores, fixes, and a rewritten section, with Taplio used to read grounding data. The workflow additionally offers to invoke update_draft or create_draft, introducing persistent changes to the user's Taplio account that are not necessary to deliver the stated audit. Offering this optionally reduces the risk, but the write capability remains outside the manifest's described scope.

Content

No source excerpt is available for this finding.

The trigger list includes everyday phrases without an explicit LinkedIn-draft prerequisite.

Low
Category
Trigger Abuse
Confidence
88% confidence
Finding

Phrases such as "is this good ?" and "what would you change ?" apply to many unrelated requests. Although the description identifies a LinkedIn-post auditing purpose, the activation instructions do not require that a LinkedIn draft be established before these generic phrases trigger the skill.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.