Back to skill

Security audit

app-ui-mockups

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marked this skill suspicious because its instructions include unpinned package commands that could run changed or compromised releases with local permissions.

Review the package commands before installing. Prefer exact, reviewed package versions and a lockfile-managed installation, and consider running the tools without access to credentials or unrelated files. The scan was partial: 31 of 32 components were scanned, with one parser limitation and binary files outside analyzer scope, so the evidence is incomplete.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (17)

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The documented npx invocation can download and execute an unpinned HyperFrames release with the agent's local permissions. The compatibility label '0.8' does not constrain package resolution; this is a conditional supply-chain risk, not evidence of a malicious package or an MCP server.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The catalog command executes an unpinned npm CLI even though its apparent purpose is only component discovery. A compromised newly resolved release could execute code with the agent's permissions; the relevant issue is package supply-chain exposure, not MCP configuration.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · assets/presets/agent-terminal-session.json (reported line 240)May include surrounding context.

json
{
      "n": 20,
      "name": "Anthropic brand-guidelines skill",
      "url": "https://github.com/anthropics/skills/blob/main/skills/brand-guidelines/SKILL.md"
    },
    {
      "n": 62,

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/agent-terminal-session.md (reported line 60)May include surrounding context.

md
[17] Ghostty source — https://github.com/ghostty-org/ghostty
[18] Claude Code docs, Interactive mode — https://code.claude.com/docs/en/interactive-mode
[19] Claude Code docs, Settings (spinner + tip line) — https://code.claude.com/docs/en/settings
[20] Anthropic brand-guidelines skill — https://github.com/anthropics/skills/blob/main/skills/brand-guidelines/SKILL.md
[49] Google Fonts metadata — https://github.com/google/fonts
[62] HyperFrames notes-typing — https://github.com/heygen-com/hyperframes/tree/main/registry/components/notes-typing
[63] HyperFrames cursor-zoom-follow — https://github.com/heygen-com/hyperframes/tree/main/registry/components/cursor-zoom-follow

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/sources.md (reported line 36)May include surrounding context.

md
Interactive mode | https://code.claude.com/docs/en/interactive-mode | Shift+Tab cycles permission modes (Manual, acceptEdits, plan…); Ctrl+T task list shows up to five tasks; "Esc to cancel" style status lines | agent-terminal-session |
| 19 | Claude Code docs, Settings | https://code.claude.com/docs/en/settings | While Claude works it shows a spinner with a one-line tip under it | agent-terminal-session |
| 20 | Anthropic brand-guidelines skill | https://github.com/anthropics/skills/blob/main/skills/brand-guidelines/SKILL.md | Dark #141413, Light #faf9f5, Orange #d97757, Blue #6a9bcc, Green #788c5d | agent-terminal-session |
| 21 | VS Code theme-defaults: 2026-dark.json, 2026-light.json, dark_plus.json (MIT) | https://github.com/microsoft/vscode/tree/main/extensions/theme-defaults/themes | Exact workbench and token colors for Dark 2026 / Light 2026 | code-editor-walkthrough |
| 22 | Visual Studio Code brand guidelines | https://code.visualstudio.com/brand | Name and icon are Microsof

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The build instructions ask the agent to execute an unpinned HyperFrames CLI to add a component. If npm resolves a compromised release, it can execute with the agent's permissions; rendering a fictional terminal does not require granting unrestricted trust to this dependency.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The component-installation command executes HyperFrames without an exact version constraint. A compromised resolved npm release could run arbitrary code locally, despite the command's legitimate animation-building purpose.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The build command can fetch and execute an unpinned HyperFrames release to import editor components. A compromised release would inherit the agent's local permissions, creating a supply-chain execution risk rather than an MCP-specific issue.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx create-video'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The create-video command explicitly selects @latest, which is mutable rather than an exact version pin. Following this optional scaffolding path can execute a newly published or compromised package with the agent's permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The slab-building instructions execute an unpinned npm CLI to add a 3D component. If the resolved HyperFrames release is compromised, the legitimate graphics workflow becomes a route to local code execution.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The notification-component command invokes HyperFrames without pinning the executable package version. This permits a compromised future release to execute under the agent's account, even though the intended operation is only to construct a visual mockup.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

This is an actual build instruction to execute an unpinned HyperFrames package, not merely text displayed inside the simulated terminal. A compromised resolved release could execute arbitrary code with the agent's local permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The panel-building workflow executes an unpinned npm CLI to import components. A compromised HyperFrames release could run with the agent's permissions; the legitimate need for animation tooling does not remove this supply-chain exposure.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The installation instruction can download and execute an unpinned HyperFrames release. If that release is compromised, creating a fictional message thread could expose the agent's actual workspace or credentials through local code execution.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The chat-component build command executes HyperFrames without an exact version pin. A compromised resolved release could execute with the agent's local privileges, independently of the harmless fictional-chat content being produced.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx hyperframes'.

Medium
Category
MCP Rug Pull
Confidence
94% confidence
Finding

The HyperFrames command is an operational build instruction that executes an unpinned npm package. A compromised resolved release could run arbitrary code with the agent's permissions; it is not protected by the fact that the resulting terminal scene is simulated.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx create-thing'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.