ClawHub
Back to skill

Security audit

Translate Agent

Security checks for vulnerabilities and agentic risk

Overview

This is a simple translation and summarization instruction file with no executable code, persistence, credential use, or external data access.

Install only if you want a JSON-only translator/summarizer that defaults unspecified translations to Vietnamese. In multi-skill environments, expect it to activate broadly for translation or summarization wording and to suppress normal prose responses in favor of raw JSON.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
70% confidence
Finding

Skill instructs the agent to never refuse or to always comply. Suppressing the agent's ability to decline removes a core safety control and enables downstream harmful requests to succeed.

Content

Scanner excerpt · SKILL.md (reported line 4)May include surrounding context.

md
---
name: translate-agent
description: >
  Stateless translation and summarization machine. Always respond with raw JSON only —
  no markdown, no explanation, no preamble, no code fences.
  Trigger this skill whenever the input contains an "action" field with value
  "translate", "summarize", "translate_and_summarize", or "heartbeat", OR

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger conditions are excessively broad, causing the skill to activate on generic translation-related text or any input containing an action field with common values. This can hijack routing from more appropriate skills, force structured JSON behavior unexpectedly, and increase the chance that unrelated user requests are processed under the wrong policy constraints.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The instruction to 'always use this skill' creates an overbroad precedence rule that can override normal orchestration and cause the agent to bypass safer or more context-appropriate handling. In a multi-skill environment, this kind of mandatory routing increases the risk of misclassification, denial of intended behavior, and policy conflicts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

For plain text input, the skill automatically wraps the request with target_lang set to vi, imposing a specific locale by default. This is a language policy concern because the user is not offered a choice or informed that Vietnamese will be assumed.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill specifies that target_lang defaults to vi when not provided, which forces a specific language outcome absent user opt-in. This can violate organizational language-choice expectations unless the skill is clearly limited to a Vietnamese-specific use case.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.