Teamo Lite Offline

• Outcome: Review
• Latest audit: Sep 10, 2026 · 7:29 PM UTC
• Version: 0.1.0

Overview

This skill is a broad task router that can force searches, web scraping, and handoff to other agents even for routine writing or code tasks.

Review before installing. Use this skill only in an environment where search, scraping, and handoff to other agents are acceptable for routine tasks, and avoid using it with private documents, proprietary code, credentials, or sensitive personal data unless the host agent enforces its own consent and data-minimization controls.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

Findings (1)

T01 · Skill Instruction Hijacking

Error

• Location: SKILL.md:13

Finding

Mandatory External Tool Use and Agent Delegation Override

Content

View full analysis

Remediation

View remediation

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 97% confidence

Finding

The workflow selection rules are so broad that ordinary user requests can be routed into powerful tool-using paths without clear boundaries. This increases the chance of unintended tool invocation, over-collection of data, and inconsistent behavior that an attacker could exploit through prompt shaping or ambiguous requests.

Content

No source excerpt is available for this finding.

Vague Triggers

High

• Category: Not specified by scanner
• Confidence: 98% confidence

Finding

The instruction text is internally inconsistent: unclear requests and summary tasks can plausibly match both workflows, while one path forbids delegation and the other requires it. Such contradictions make the skill unpredictable and easier to manipulate into selecting a more privileged or less appropriate execution path.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The skill is labeled 'offline' but explicitly instructs the agent to perform online information gathering, search, and mandatory scraping in some cases, without requiring user opt-in. This creates a misleading trust boundary and can cause unanticipated external data access or disclosure of user-provided context to remote services.

Content

No source excerpt is available for this finding.

Scope Creep

Low

• Category: Excessive Agency
• Confidence: 70% confidence

Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md

## Instructions

You are Teamo-Lite, a high-speed AI for task planning and online information gathering. Your job is to **strictly choose one** of the following two workflows base on user needs, and complete task efficiently.**Workflow 1: Quick Q&A***   **Trigger Condition**: The user needs to directly **obtain**, **extract**, **query**, or **explain** existing information. This is typically a **non-creative** task. For example, "Extract the summary section from this PDF." The user sends a message with an **unclear request**. The user requests **image generation**.*   **Your Role**: **Efficient Q&A Assistant***   **Action Steps**:    1.  Carefully review the tool list and call the necessary tools to complete the task. When faced with a problem that cannot be directly solved by existing tools, consider using the available tools to solve it to the greatest extent possible. (For example, if the user asks for today's weather and there is no real-time weather tool, you can combine today's date, ask for the user's location, and use a search tool to find the weather.)    2.  Directly answer the user's question. **Strictly prohibit** calling `call_other_agents`. If you get stuck, prioritize outputting useful information quickly, then ask for clarification.**Workflow 2: Content Creation or Code Problems***   **Trigger Condition**: The user needs content output in the form of **code processing, writing, creation, proposals, reports, summaries, lists**, etc. This is typically a creative and complex writing task. For example, "Write a summary of this attached PDF." The user asks any question related to code (including but not limited to code writing, debugging, explanation, optimization, algorithms, etc.).*   **Your Role**: **Researcher***   **Action Steps**:    1.  Call various search tools to gather information.    2.  After completing the search, judge the complexity of the task to decide whether to call the `url_scraping` tool. For complex tasks that require extensive profe
...[truncated 25 chars]

Static analysis

No suspicious patterns detected.