Back to skill

Security audit

Agent Master Skills Clawhub Pkg

Security checks for vulnerabilities and agentic risk

Overview

The scanners classify this skill as suspicious because its catalog descriptions may trigger unrelated skill use, though they report no executable scripts or confirmed malicious behavior.

Consider whether you want a skill catalog whose descriptions may lead an agent to invoke other skills during unrelated tasks. The scans were complete for the three reported Markdown components, and no executable scripts were reported; the suspicious verdict is driven by instruction scope, not a finding of malicious intent.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (5)

Agent Config Directory Access

High
Category
Agent Snooping
Confidence
90% confidence
Finding

Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.

Content

Scanner excerpt · README.md (reported line 65)May include surrounding context.

md
| **Codex CLI** / Replit Agent | `.agents/skills/` | `~/.agents/skills/` |
| **Cursor** | `.cursor/skills/` | `~/.cursor/skills/` |
| **Gemini CLI** | `.gemini/skills/` | — |
| **Google Antigravity** | `.agent/skills/` | `~/.gemini/config/skills/` |
| **Hermes Agent** | `.hermes/skills/` | `${HERMES_HOME:-$HOME/.hermes}/skills/` |
| **Grok Build** | `.grok/skills/` | `~/.grok/skills/` |
| **Devin** | `.devin/skills/` | `${XDG_CONFIG_HOME:-$HOME/.config}/devin/skills/` |

The byterover description requires activation before any work, without limiting it to knowledge-management tasks.

Medium
Category
Trigger Abuse
Confidence
98% confidence
Finding

The description instructs the agent to use byterover before any work, extending its authority beyond relevant knowledge-management tasks. Although quoted in a catalog, this unconditional instruction can influence an agent consuming the index and cause unrelated tool invocation; the supplied content does not establish actual data exfiltration.

Content

No source excerpt is available for this finding.

The find-skills description includes the everyday question pattern 'how do I do X' as an activation condition.

Medium
Category
Trigger Abuse
Confidence
94% confidence
Finding

L432 associates skill discovery and installation with generic how-to questions. Such questions commonly request direct guidance rather than an additional skill, so this trigger can activate discovery unintentionally.

Content

No source excerpt is available for this finding.

The using-superpowers description activates on the start of any conversation.

Medium
Category
Trigger Abuse
Confidence
98% confidence
Finding

The description claims activation at the start of any conversation and requires use of a skill tool, regardless of the user's task. When consumed by an agent, this catalog entry can redirect unrelated workflows into another skill without a task-specific reason, although no destructive behavior is demonstrated here.

Content

No source excerpt is available for this finding.

The skill description spans many unrelated domains without specifying when to invoke this catalog.

Medium
Category
Trigger Abuse
Confidence
88% confidence
Finding

The description names broad areas such as productivity, marketing, frontend, and DevOps but does not limit activation to requests to discover or browse skills. This leaves it unclear whether ordinary tasks in those areas should activate the catalog.

Content

No source excerpt is available for this finding.