ClawHub
Back to skill

Security audit

Scripting And Storyboarding

Security checks for vulnerabilities and agentic risk

Overview

This is a planning-only video pre-production skill with clear human approval boundaries and no executable install or persistence behavior.

Installers should understand that this skill helps draft video production plans and AI previz briefs. Review its outputs for consent, disclosure, originality, and feasibility before filming, generating, editing, or publishing.

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
  • 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 (2)

Autonomous Decision Making

Medium
Category
Excessive Agency
Content
(shot table or AI-previz briefs), the numbered shot list, the batch-shoot plan, the timing pass, and the
  footage-labeling map. The **human** shoots (or generates), judges every take/clip, and approves — the agent
  cannot see footage, judge a performance, or operate a camera, and never fabricates "that take works."
- **AI previz frames** follow the image rules (`flux`/`image-prompt`): no real-person likeness without
  consent, character bibles built on original/consented characters, AI-disclosure where the frames themselves
  get published.
- **Production integrity:** no staged-as-candid content (scripted actors posing as unaffiliated strangers =
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.

Vague Triggers

Low
Confidence
89% confidence
Finding
This is a manifest-style JSON file, so vague-trigger review applies. The description explains the skill's purpose in broad natural language ('turn an idea into a shootable, editable plan') but does not define specific activation phrases, scope limits, or negative examples, which could make invocation boundaries unclear in systems that route by description.

Static analysis

No suspicious patterns detected.