Back to skill

Security audit

TED Talks

Security checks for vulnerabilities and agentic risk

Overview

This skill is labeled as TED Talks but its instructions mostly describe an unrelated paid education-course platform, so users should review it before installing.

Install only if you are comfortable with a skill whose branding and documented behavior do not match. The artifact does not show code execution or data access, but it should be renamed or rewritten so the TED Talks purpose, examples, and capabilities are consistent before normal use.

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
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The documented behavior materially conflicts with the declared skill purpose: instead of TED talks, bilingual subtitles, and insight summaries, it describes an education-course platform with exams, subscriptions, certificates, and mock scores. This mismatch can misroute user requests, trigger the skill on unrelated education queries, and misrepresent capabilities, which is dangerous because users and orchestrators may trust the advertised identity while the actual prompt steers into a different domain.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The description and content specify Chinese-language outputs and bilingual subtitles, but there is no indication that users can choose their preferred language or locale. This can violate language/locale policy when a skill implicitly enforces one language without opt-in or a documented regional requirement.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The title and invocation examples present the skill as 'TED Talks,' but the body documents unrelated learning-platform functions. This identity contradiction increases the chance of deceptive routing or accidental invocation under a trusted brand name, causing users to disclose intent or rely on outputs from a skill that does not do what it claims.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The example trigger phrases are very broad, such as asking how to use the skill for a 'specific scenario task' or for 'latest玩法/优惠信息,' which can overlap with ordinary conversation and unrelated shopping or help requests. Overbroad triggers increase unintended activation risk, especially when combined with the skill's already-misaligned identity, leading to prompt hijacking of user flows or irrelevant skill selection.

Content

No source excerpt is available for this finding.