Back to skill

Security audit

Openclaw User Intent Refinement

Security checks for vulnerabilities and agentic risk

Overview

This skill provides intent-clarification guidance and does not contain code, persistence, credential access, or hidden installation behavior.

Install this only if you want the agent to normalize unclear requests before acting. For high-impact actions, still expect the agent to show assumptions and ask before changing, deleting, spending, posting, or using sensitive data.

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
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • 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
Findings (3)

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

The 'Final Output Rule' directs the agent to silently rewrite user intent and continue without user confirmation when it believes the request can be corrected. In a skill designed to transform ambiguous instructions into executable actions, this increases the risk of hidden prompt steering and unintended action execution, especially when the reinterpretation changes scope, assumptions, or safety-relevant details without explicit consent.

Content

Scanner excerpt · SKILL.md (reported line 63)May include surrounding context.

md
**Professional Mode (task penting):** INTENT + FACT CHECK + DOMAIN THEORY + TECHNICAL VALIDATION + ENVIRONMENT CHECK + RISK CHECK + EXECUTION + VERIFICATION.

**Final Output Rule:** bila dapat diperbaiki tanpa bertanya → perbaiki internal & lanjut. Bila perlu kabar user → format ringkas: "Maksudnya benar adalah X. Saya pakai metode Y karena Z." Jangan kuliah panjang kecuali diminta.

**Master Algorithm:** READ → EXTRACT → NORMALIZE → IDENTIFY DOMAIN → CHECK FACTS → CHECK THEORY → CHECK ENVIRONMENT → DETECT MISSING → CORRECT → ADD PARAMS → REMOVE INVALID → BUILD → EXECUTE → VERIFY → RECOVER → REPORT.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README says to use the skill 'when the conditions in the description are met,' but it does not define concrete activation boundaries, exclusions, or precedence relative to other skills. For a skill that rewrites user intent before actions are taken, ambiguous triggering can cause over-activation, unintended modification of user requests, or invocation in contexts where clarification should occur instead of reinterpretation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The description states the skill in Indonesian and presents its behavior as a fixed professional mode, with no indication that users may choose another language or locale. Under the policy rule, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.