ClawHub
Back to skill

Security audit

Consilium

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed multi-model advisory workflow that sends the user's question to chosen model providers and does not include hidden code, destructive actions, or unrelated authority.

Before installing, understand that your prompt will be sent to several configured model providers and may be stored according to normal OpenClaw session settings. Use the panel only with providers you trust, and rerun model selection with --models if you want to control which models receive your questions.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

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 12)May include surrounding context.

md
**Unlike other council skills:** this uses genuinely different models (Anthropic + OpenAI + Google + others), not one model playing multiple roles. Different training data = different blind spots = better coverage.

**Always respond in the same language as the user's question.**

## Examples

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest says to use the skill when the user 'asks for multi-model analysis of a decision, architecture, strategy, or any complex problem.' The phrase 'any complex problem' is very broad and lacks limiting conditions or negative examples, which could cause unintended invocation for ordinary advisory requests.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The instruction requires the skill to detect the user's question language and then force a strict response language, explicitly discouraging more flexible phrasing. This is a natural-language policy concern because it imposes a language/locale behavior automatically rather than offering the user a choice or documenting opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

This markdown file includes the natural-language feature "Explicit language detection," which may imply locale or language handling is enforced automatically. Because the roadmap does not mention user choice, opt-in, or a justified region-specific constraint, it risks conflicting with the policy against forcing language or locale behavior without user consent.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.