Back to skill

Security audit

Haqq Workbench

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent workflow router, but it broadly routes agents toward public posting, ClawHub publishing, browsing, and persistent memory without clear confirmation gates.

Install only if you want an Arabic Haqq workflow router that can steer your agent toward other skills. Before using it for publishing, messaging, ClawHub releases, browsing, or memory storage, require an explicit preview and confirmation for each external or persistent action, and verify account/platform scope yourself.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (9)

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

Publishing and messaging skills are mapped to extremely generic phrases like 'publish', 'send', or 'interact', which are likely to appear in ordinary conversation. If these triggers activate automatically, the agent could post content, send messages, or engage on external platforms without sufficiently explicit user authorization, making this materially more dangerous than benign lookup skills.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill advertises itself as a "MASTER skill" for broadly defined tasks like creating content, publishing, verification, and any task requiring multiple skills. That activation scope is so wide that an agent could invoke it for many ordinary workflows, increasing the chance of unnecessary tool orchestration and accidental access to publishing or memory-related capabilities without clear user intent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The publishing workflow instructs the agent to schedule, publish, and monitor replies across external platforms, but it does not require an explicit confirmation step or warn that these actions affect public accounts and external data. In practice, this can lead to unintended posting, reputational harm, or disclosure of unreviewed religious or factual content if the orchestration is triggered too eagerly.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill-creation workflow includes automatic publication to ClawHub and announcement on platforms without an explicit warning or approval checkpoint. That creates a direct path from generating a new skill to externally releasing it, which is risky because flawed, unsafe, or sensitive content could be published before review.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger phrases in this section are broad, common Arabic terms such as requests to search or verify, so normal user messages could unintentionally invoke higher-privilege verification or research skills. This creates routing ambiguity and can cause unnecessary tool use, data exposure to external services, or execution of workflows the user did not explicitly request.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The memory and browser triggers are also too general, using ordinary phrases like 'remember', 'save', 'open site', or 'browse', which can easily occur in routine dialogue. In context, this is risky because memory writes can persist unintended data and browser activation can initiate external navigation or retrieval without a clearly bounded user request.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This code returns a fixed Arabic message when no task is detected, regardless of the user's preferred language. The policy requires avoiding forced language or locale choices unless the skill offers opt-in or clearly documents a justified locale restriction.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The command-line interface prints workflow labels and descriptions in Arabic only, with no mechanism for user language selection. This creates a natural-language locale constraint that is not optional or justified in the file.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

Aside from the manifest metadata line, the operational instructions and policy text are written in Arabic and do not indicate that users may choose another language. This can constitute a language policy issue when a specific language is imposed without explicit opt-in or documented regional justification.

Content

No source excerpt is available for this finding.