Back to skill

Security audit

stealth-break

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a disclosed macOS break-reminder helper, but it asks agents to create persistent cron-based notifications without enough scoping, removal guidance, or clear containment.

Install only if you are comfortable with a skill that may add cron-based macOS reminder jobs. Before allowing setup, ask to see the exact cron entry and osascript command, confirm the frequency, and make sure you know how to list and remove the cron job later. Prefer a pinned install command from a trusted source.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
README.md:39
Finding

Unpinned Package Execution in Installation Instructions

Content
View full analysis

Vulnerability Details

File Location: README.md, lines 39-41
Vulnerability Type: Unpinned third-party package execution
Risk Level: Medium

Vulnerable Code

bash
npx clawhub install stealth-break

Technical Analysis

The documented installation command invokes clawhub through npx without specifying an exact package version or integrity constraint. If the package is not already available locally, npx may download and execute the version currently resolved by the configured package registry.

Because package resolution is mutable, the code executed during installation can differ from the version that existed when this Skill was reviewed. Compromise of the package, its publisher account, the registry resolution path, or a newly published malicious version could therefore introduce arbitrary behavior through the CLI itself or its package lifecycle mechanisms.

The repository does not itself contain malicious dependency code, and exploitation depends on compromise or unexpected modification of the externally resolved package.

Attack Path

  1. An attacker compromises the clawhub package, its publisher account, or an applicable package-resolution source.
  2. The attacker publishes a malicious version that remains compatible with the unversioned package name.
  3. A user follows the README and runs npx clawhub install stealth-break.
  4. npx resolves and downloads the attacker-controlled package version.
  5. The downloaded CLI or associated lifecycle code executes with the privileges of the user running the installation command.
  6. The malicious package can access or modify resources available to that user.

Impact Assessment

Successful exploitation could result in arbitrary code execution under the installing user's account. The reachable scope includes user-readable files, user-writable files, environment variables, credentials exposed to the process, and network resources accessible ...[truncated 251 chars]

Remediation
View remediation

Remediation Suggestions

  1. Pin clawhub to an exact, reviewed version in the installation command, for example:
    bash
    npx clawhub@<reviewed-exact-version> install stealth-break
    
  2. Do not use a floating tag such as latest, a caret range, or a tilde range for security-sensitive installation workflows.
  3. Document the expected registry and verified package publisher so users can validate package provenance.
  4. Where supported, verify the downloaded package against a trusted checksum, signature, or registry provenance attestation before execution.
  5. Review the pinned package version, including lifecycle scripts and transitive dependencies, before recommending it.
  6. Execute installation with a least-privileged user and avoid exposing unrelated credentials or sensitive environment variables to the process.
  7. Update the pinned version only through a controlled review process.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (9)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The listed trigger phrases are broad, conversational, and likely to appear during ordinary user dialogue, which can cause the skill to activate unintentionally. Because this skill is described as setting reminders and creating cron-based persistence, accidental invocation could lead to unwanted configuration changes or recurring notifications without clear user intent.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The README instructs users to run npx clawhub install stealth-break without pinning a specific version, which means installation behavior depends on the latest package published at execution time. This creates a supply-chain risk: a compromised or malicious upstream package version could be fetched and run unexpectedly on the user's machine.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README says the skill will automatically create cron tasks and send macOS system notifications, but it does not clearly warn users that this introduces persistent system changes. Silent or poorly disclosed persistence mechanisms are risky because they can continue executing after initial use, surprise users, and be abused to normalize behavior that resembles persistence often seen in unwanted software.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation triggers are broad, everyday phrases such as being tired or needing to relax, which can cause the skill to activate unintentionally in many normal conversations. Because the skill can lead to command execution and persistent scheduling later, overbroad invocation increases the chance of users being guided into system-affecting actions without a clearly intentional opt-in.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The skill metadata and description frame the capability as macOS notification delivery, but the body expands this into persistent cron-based automation. This capability mismatch is dangerous because users and policy layers may underestimate the level of system access and persistence the skill will exercise.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill directs the agent to create persistent cron jobs, which modifies the user's system state beyond a one-shot notification action. Even though reminders are part of the stated purpose, persistence creates an automation foothold that can surprise users, remain after the session ends, and be repurposed or misconfigured without explicit consent and transparency.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill tells the agent to use cron and exec/osascript without a clear warning that this will execute commands and install persistent scheduled tasks. Lack of disclosure undermines informed consent and makes it easier for a seemingly wellness-oriented skill to perform system modifications users did not fully anticipate.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The README presents the skill name, description, and trigger phrases entirely in Chinese, implying a fixed language for use and invocation. There is no indication that other languages are supported or that the user can opt into a preferred locale.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

The package description is written entirely in Chinese, which imposes a single language in user-facing metadata without indicating any language choice or opt-in. The file does not document that this skill is intended only for a Chinese-speaking audience or a region-specific use case.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.