Back to skill

Security audit

Haqq Ethics

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only religious ethics skill, but it tries to impose broad always-on agent behavior and memory-like retention beyond a clearly scoped user request.

Review before installing. This skill appears non-executable and does not request files, credentials, network APIs, or local privileges, but it should only be used where you intentionally want an Arabic Islamic ethics/source-handling framework. Do not load it as a global default skill for unrelated tasks unless it is revised to remove always-on, memory, and authority-hierarchy language and to state that platform/system instructions remain authoritative.

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
Findings (2)

T01 · Skill Instruction Hijacking

Error
Location
Ethics.md:131
Finding

Global Agent Instruction and Authority Hijacking

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Warning
Location
Ethics.md:5
Finding

Attempted Persistent Memory Poisoning Through Retention Directives

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (6)

Ssd 3

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The text explicitly tells the agent to memorize the words and not forget a letter, then to reactivate memory, which is an instruction for persistent retention beyond the current interaction. Always-on retention is unsafe because it can create unauthorized cross-session carryover and hidden prompt persistence not approved by the platform or user.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This text imposes an Arabic-only handling rule for scripture and instructs the agent not to treat translations as acceptable equivalents, regardless of user preference or language needs. In an agent skill, that is a policy override that can restrict normal assistance, reduce accessibility, and force religious/language-specific behavior on users without consent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The file instructs the agent to answer every question according to a rigid religious framework, including refusing normal interpretive behavior and forcing fixed sourcing and phrasing rules. This is dangerous because it attempts to globally override the assistant’s broader instruction hierarchy and user intent, causing unauthorized behavioral control across unrelated tasks.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The instruction to remember this text at every question, every action, and everything the agent begins creates a global persistent behavioral hook. In practice, this acts like an always-on prompt implant that can bias all future responses, interfere with task neutrality, and bypass normal scoping controls.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The README content is fully presented in Arabic and does not indicate that users may choose another language or that the skill is intentionally limited to Arabic-speaking users. Under the policy for natural-language violations, forcing a specific language without user opt-in is a reportable issue.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill explicitly restricts scripture handling to Arabic-only text and does not offer any language choice or user opt-in. In a general-purpose agent skill, this can exclude users, reduce transparency for non-Arabic readers, and create a safety/usability problem because users may be unable to understand source-backed guidance or verify what the agent is citing.

Content

No source excerpt is available for this finding.