Back to skill

Security audit

myreels-storyboard

Security checks for vulnerabilities and agentic risk

Overview

This skill is a storyboard planning aid that writes local project files and metadata, with no executable code, hidden payloads, credential access, or destructive behavior found.

Before installing, understand that this skill can organize storyboard work into local folders and may save story briefs, prompts, request payloads, task IDs, metadata, returned URLs, and generated media references. Use a deliberate output directory and avoid using sensitive or confidential story material unless local retention is acceptable.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (10)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Broad activation phrases like script, storyboard, or short drama increase the chance of accidental activation in ordinary conversation. If triggered unintentionally, the skill may begin collecting project details, drafting outputs, or preparing workspace actions outside the user’s intent, which is risky given its file-writing behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documentation tells users this skill only handles storyboard design, yet later workflow steps instruct it to manage downstream myreels-api artifacts and rerun state. That creates a deceptive trust boundary: users may invoke the skill expecting a planning-only assistant, while it is actually authorized to organize and persist generation-related data on disk.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% 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 · SKILL.md (reported line 50)May include surrounding context.

md
After collecting inputs:
- Use the current OpenClaw session's default LLM to understand the story semantically
- Infer core roles, recurring relationships, and phase-by-phase visual beats from the user's actual plot, not only from surface keywords
- Do not ask the user to configure an extra LLM just to use this skill
- Do not rely on a bundled local storyboard script for core planning; OpenClaw should draft the CSVs directly
- The old helper `scripts/generate_storyboard.py` is deprecated and should not be referenced in this workflow

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill’s manifest and early documentation state it only performs storyboard planning and should hand off actual media generation to myreels-api, but later sections direct it to manage generation outputs, request payloads, task snapshots, metadata, and reruns. This mismatch expands the skill’s effective scope and can cause users or orchestrators to grant it file-system access and generation-adjacent authority they did not expect.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs automatic creation of a project workspace and persistence of drafts, approvals, images, videos, request payloads, and metadata in the user’s current working directory or requested output directory, without a clear up-front warning or consent step. Unannounced file writes can leak sensitive story content, clutter repositories, overwrite expectations about workspace safety, and expose prompt, task, or media metadata that users did not realize would be stored.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

Line L096 states 'Use English for the actual generation prompt' as a blanket instruction. This forces a specific language for operation without user opt-in and does not document a justified region- or tool-specific requirement.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest says this skill focuses on storyboard design/content planning and that actual image/video generation requires the separate myreels-api skill. However, this file specifies a full project workspace for generated images/videos, request payloads, task polling snapshots, metadata, reruns, and delivery artifacts, which goes beyond mere storyboard design guidance into generation operations management.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The workspace and reproducibility rules instruct the system to persist extensive user content, prompts, metadata, URLs, notes, and generation records across many files. If the stories contain sensitive personal, proprietary, or confidential material, this creates a real data-retention and inadvertent disclosure risk, especially because the guidance encourages broad, long-lived storage of exact payloads and outputs rather than data minimization.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

Line L130 includes Chinese text '(上下)' inside an otherwise English document, which imposes a specific language/localization choice without explaining or offering an alternative. This is a natural-language locale inconsistency that could conflict with organizational language policy requiring user choice or clear justification.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The manifest workflow describes storyboarding and then handing off to myreels-api for generation, but this reference file also instructs output formatting for Feishu Bitable and Telegram attachments/text tables. Those delivery/integration behaviors are not reflected in the manifest's declared scope.

Content

No source excerpt is available for this finding.