ClawHub
Back to skill

Security audit

spec-writing

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only specification-writing skill with some broad triggers and documentation inconsistencies, but no hidden execution, credential access, persistence, or destructive behavior.

Before installing, be aware that this skill may trigger on ordinary requirements or specification discussions and its templates contain minor consistency issues. It appears safe as a writing aid, but users should review generated specs for explicit assumptions and measurable success criteria.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger list includes very broad generic terms such as "specification," "requirements," and "acceptance-criteria," which can cause the skill to activate in many ordinary conversations unrelated to an explicit request for this capability. Over-broad activation increases the chance of unintended skill invocation, context hijacking, and interference with other more appropriate skills or core agent behavior.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The activation condition "starting a new feature and needing a spec before planning" is a broad natural-language situation rather than a precise invocation rule. This makes routing subjective and easier to trigger accidentally from normal project discussion, which can lead to unexpected behavior, misapplied instructions, or skill precedence issues.

Intent-Code Divergence

Medium
Confidence
96% confidence
Finding
Assumptions is presented as required content in the mandatory sections area but later also treated as optional, creating ambiguity about whether specs must document unverified conditions. In this skill context, that can lead users to omit critical assumptions, increasing the chance of misunderstood scope, invalid requirements, or unsafe downstream decisions based on hidden premises.

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
**What to Include**:
- Conditions assumed true that, if proven false, would
  invalidate parts of this specification
- Scope boundaries accepted without verification
- Environmental or organizational givens

**Template**:
Confidence
75% confidence
Finding
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The document gives contradictory instructions by declaring Success Criteria mandatory earlier and then describing it again as optional later. In a spec-writing skill, inconsistent structure guidance can cause incomplete or inconsistent specifications, which may omit measurable outcomes and weaken downstream planning, testing, or compliance review.

Static analysis

No suspicious patterns detected.