Back to skill

Security audit

Generate Personal Brand Ad Creative Brief

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple advertising-brief helper, but it asks for shell access that is not explained or needed for its stated workflow.

Review this skill before installing because it grants Bash access despite only needing creative-brief generation. If you use it, prefer a version that removes Bash or run it in a restricted sandbox and avoid giving it sensitive files or credentials.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (2)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:9
Finding

Unnecessary Bash Permission Violates Least-Privilege Boundaries

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 9
Vulnerability Type: Excessive tool permission
Risk Level: Medium

Vulnerable Code:

yaml
allowed-tools: Bash, Read

Technical Analysis

The Skill grants access to Bash, although its documented workflow is limited to clarifying advertising requirements, generating a creative brief, enriching assets, and refining output. No documented step requires operating-system command execution.

Granting Bash access expands the Skill's authority beyond its stated business purpose. If malicious, misleading, or otherwise untrusted task content influences the agent, the shell permission may provide a channel for executing commands on the host. The practical severity depends on the runtime's sandboxing, user privileges, command-approval controls, and network restrictions.

Attack Path

  1. A user loads the Skill, granting it the tools declared in allowed-tools.
  2. The agent receives adversarial or compromised campaign content containing instructions to invoke shell commands.
  3. The agent uses the unnecessarily available Bash tool.
  4. Commands execute with the privileges and environmental access of the agent runtime.
  5. Depending on runtime controls, the commands could read accessible files, modify workspace data, launch processes, or initiate network requests.

Impact Assessment

Successful exploitation could expose any files, environment data, commands, and network resources available to the agent's operating-system identity. It could also permit modification or deletion of writable data. This project does not itself contain a malicious command or demonstrate privilege escalation to an administrator account; the risk arises from excessive capability combined with potentially untrusted future input.

Remediation
View remediation

Remediation Suggestions

  • Remove Bash from allowed-tools because it is not required by the documented workflow.
  • Grant only the minimum tools necessary for text and image generation.
  • If shell access becomes necessary for a future feature, replace unrestricted Bash access with narrowly scoped operations and an explicit command allowlist.
  • Run any required command in a sandbox with restricted filesystem access, blocked outbound networking by default, low operating-system privileges, and execution time limits.
  • Require explicit user confirmation before commands that write files, access sensitive locations, install software, or use the network.

T08 · Insecure Dependencies

Warning
Location
README.md:8
Finding

Installation Instructions Use Mutable and Unverified Upstream Sources

Content
View full analysis

Vulnerability Details

File Location: README.md, lines 8 and 13-14
Vulnerability Type: Unpinned supply-chain source
Risk Level: Medium

Vulnerable Code:

bash
clawhub install qiaomu-generate-personal-brand-ad-creative-brief
bash
git clone https://github.com/qiaomucom/generate-personal-brand-ad-creative-brief.git
cp -r generate-personal-brand-ad-creative-brief ~/.openclaw/skills/generate-personal-brand-ad-creative-brief

Technical Analysis

Both documented installation methods retrieve mutable upstream content without selecting an immutable version or commit. The instructions also provide no checksum or signature verification.

Consequently, the content installed by a future user may differ from the content reviewed in this audit. If the registry package, source repository, maintainer account, or release process is compromised, modified Skill instructions or additional files could be delivered under the expected package name.

The reviewed files do not prove that either current upstream source is malicious. This finding concerns the absence of reproducibility and integrity controls in the documented installation process.

Attack Path

  1. An attacker compromises an upstream publisher account, registry entry, repository, or release workflow.
  2. The attacker replaces or modifies the mutable package or default Git branch.
  3. A user runs one of the documented unpinned installation commands.
  4. The altered content is downloaded without checking a trusted signature or expected digest.
  5. The content is installed in ~/.openclaw/skills/.
  6. When the modified Skill is loaded, attacker-controlled instructions or components operate with the tools and permissions granted to that Skill.

Impact Assessment

The resulting privileges depend on the contents of the compromised package and the OpenClaw runtime configuration. A malicious replacement could alter agent behavior and potentia ...[truncated 174 chars]

Remediation
View remediation

Remediation Suggestions

  • Pin ClawHub installation to a specific immutable version where supported.
  • For manual installation, check out a reviewed commit hash or signed release tag rather than the repository's mutable default branch.
  • Publish a cryptographic checksum for each release and verify it before copying files into the active skills directory.
  • Sign releases or commits and document verification against a trusted maintainer key.
  • Review the complete downloaded package before activation, including any newly introduced scripts, links, tool permissions, and metadata.
  • Prefer an installation process that fails closed when signature, version, or checksum validation does not succeed.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (1)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The suggested prompt "[Generate Personal Brand Ad Creative Brief] for my team" is generic enough that it can overlap with normal user requests about ad planning, increasing the chance the skill is invoked unintentionally. In agent environments, overly broad triggers can cause incorrect tool/skill selection, leading to irrelevant actions, prompt hijacking surface expansion, or unintended use of connected capabilities such as chat or image generation.

Content

No source excerpt is available for this finding.