Back to skill

Security audit

Opus Clip

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed OpusClip workflow guide that emphasizes human review before publishing and contains no executable code or hidden persistence.

Installers should still treat it as operational guidance for social video workflows: verify current OpusClip pricing/API availability in-app, keep human approval before publishing, and do not grant connected publishing tools permission to auto-post without review.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (2)

Autonomous Decision Making

Medium
Category
Excessive Agency
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.

Content

Scanner excerpt · references/workflows-and-templates.md (reported line 42)May include surrounding context.

md
## Never
Auto-post unreviewed clips · treat the score as a guarantee or auto-delete threshold · ship a clip that
misrepresents a speaker · skip the caption accuracy pass · feed burned-in-subtitle video · leave auto-music on
for commercial posts without checking rights · process untrimmed sources · assume the editor/API is on your tier
without checking · leave exports undownloaded near renewal · claim WoopSocial clips video.

Autonomous Decision Making

Medium
Category
Excessive Agency
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.

Content

Scanner excerpt · references/workflows-and-templates.md (reported line 43)May include surrounding context.

md
## Never
Auto-post unreviewed clips · treat the score as a guarantee or auto-delete threshold · ship a clip that
misrepresents a speaker · skip the caption accuracy pass · feed burned-in-subtitle video · leave auto-music on
for commercial posts without checking rights · process untrimmed sources · assume the editor/API is on your tier
without checking · leave exports undownloaded near renewal · claim WoopSocial clips video.