Back to skill

Security audit

Credential Auditor

Security checks for vulnerabilities and agentic risk

Overview

This skill is not clearly malicious, but it advertises automated credential testing and brute-force workflows without strong per-target controls, and some advertised capabilities are incomplete.

Install only if you are a security tester working on systems you are explicitly authorized to assess. Treat the brute-force and external-tool workflows as high-risk, verify the exact version you install, prefer an isolated environment, avoid symlinking a mutable checkout, and confirm target scope before running any credential test.

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 (2)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:195
Finding

Unpinned and Unnecessary Python Dependencies Installed from a Mutable Package Index

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:195, INSTALL.md:33, INSTALL.md:117, README.md:90, README.md:450, and PROJECT_INFO.md:113
Vulnerability Type: Insecure third-party dependency installation
Risk Level: Medium

Complete Vulnerable Code Snippet:

bash
pip install paramiko requests colorama

Technical Analysis

The installation documentation directs users to install three Python packages without exact versions, integrity hashes, a lockfile, an isolated virtual environment, or an explicitly trusted package index. Consequently, the installed code can differ between installations and can change after the Skill has been reviewed.

The audited scripts do not import paramiko, requests, or colorama, making this supply-chain exposure unnecessary for the current implementation. If a package, transitive dependency, maintainer account, or configured package index is compromised, installation can introduce attacker-controlled code. Python packages may execute build backend or installation logic during installation.

Attack Path

  1. An attacker compromises a listed package, one of its transitive dependencies, its publishing account, or a package index used by the victim.
  2. The attacker publishes a malicious version that remains compatible with the unconstrained package names.
  3. A user follows the project documentation and runs:
    bash
    pip install paramiko requests colorama
    
  4. pip resolves the attacker-controlled or compromised release because no version or hash constraints are present.
  5. Malicious build or installation logic executes with the privileges of the user running pip.
  6. The installed package may subsequently execute again if imported by future versions of the Skill or by other software in the same Python environment.

Impact Assessment

Successful exploitation could execute arbitrary code with the privileges of the user performing install ...[truncated 488 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove paramiko, requests, and colorama from the installation instructions until the implementation actually requires them.
  2. If dependencies become necessary, define them in a reviewed dependency manifest using exact versions.
  3. Generate a hash-locked requirements file and require hash verification, for example:
    bash
    python3 -m pip install --require-hashes -r requirements.lock
    
  4. Install dependencies in a dedicated virtual environment rather than the global or user-wide Python environment:
    bash
    python3 -m venv .venv
    . .venv/bin/activate
    python3 -m pip install --require-hashes -r requirements.lock
    
  5. Document the trusted package index and disable unintended extra indexes to reduce dependency-confusion risk.
  6. Use automated dependency scanning and review dependency updates before regenerating the lockfile.
  7. Keep dependency declarations consistent across SKILL.md, INSTALL.md, README.md, and PROJECT_INFO.md.

T08 · Insecure Dependencies

Warning
Location
INSTALL.md:9
Finding

Skill Installation and Updates Retrieve Unpinned Content from a Mutable Git Repository

Content
View full analysis

Vulnerability Details

File Location: INSTALL.md:9, INSTALL.md:17-22, INSTALL.md:87-93, README.md:77, and README.md:84-86
Vulnerability Type: Mutable remote supply-chain dependency
Risk Level: Medium

Complete Vulnerable Code Snippets:

bash
# Method 1: Clone the repository
git clone https://github.com/jeanphorn/credential_auditor.git
cd credential_auditor
bash
# Create the OpenClaw Skill directory if it does not exist
mkdir -p ~/.openclaw/workspace/skills/

# Copy the Skill into OpenClaw
cp -r . ~/.openclaw/workspace/skills/credential_auditor/

# Alternatively, use a symbolic link
ln -s $(pwd) ~/.openclaw/workspace/skills/credential_auditor
bash
# If installed using git clone
cd ~/.openclaw/workspace/skills/credential_auditor
git pull

# If installed using a symbolic link
cd /path/to/credential_auditor
git pull

Technical Analysis

The documented installation process retrieves the current head of a remotely mutable GitHub repository and places that content in the OpenClaw Skill directory. The update instructions similarly recommend an unrestricted git pull. No reviewed release tag, immutable commit hash, cryptographic signature, checksum, or required diff review is specified.

The symbolic-link installation option further couples the active OpenClaw Skill directly to a mutable checkout. Changes pulled into that checkout can therefore alter both executable scripts and SKILL.md, which controls how the agent interprets and uses the Skill.

The repository content observed during this audit did not contain a malicious payload, instruction hijacking, persistence logic, or data-exfiltration code. The risk arises because the documented process permits future, unaudited remote content to replace the reviewed content.

Attack Path

  1. An attacker compromises the remote repository, a maintainer account, or the release workflow.
  2. The attacker com ...[truncated 1258 chars]
Remediation
View remediation

Remediation Suggestions

  1. Publish immutable, versioned release artifacts rather than instructing users to install the default branch head.
  2. Pin installation instructions to a reviewed commit hash or signed release tag.
  3. Publish a SHA-256 or stronger checksum for every release artifact and require users or the installer to verify it before installation.
  4. Sign release tags and artifacts using a documented, verifiable signing identity.
  5. Replace unrestricted git pull instructions with an update procedure that fetches a specific reviewed version and displays the diff before activation.
  6. Avoid symlinking a mutable development checkout directly into the active OpenClaw Skill directory. Copy a verified immutable artifact instead.
  7. Run static analysis and automated tests on each release before publishing it.
  8. Document rollback procedures and retain previously verified releases so users can recover safely from a compromised update.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Findings (30)

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

The uninstall command performs recursive forced deletion, which can be dangerous if the path is modified, mistyped, or unexpectedly expanded. In an install guide, users often copy/paste commands directly, so even a targeted 'rm -rf' merits caution because mistakes can lead to irreversible data loss.

Content

Scanner excerpt · INSTALL.md (reported line 101)May include surrounding context.

bash
# 删除技能目录
rm -rf ~/.openclaw/workspace/skills/credential_auditor

故障排除

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

The uninstall command performs recursive forced deletion, which can be dangerous if the path is modified, mistyped, or unexpectedly expanded. In an install guide, users often copy/paste commands directly, so even a targeted 'rm -rf' merits caution because mistakes can lead to irreversible data loss.

Content

Scanner excerpt · INSTALL.md (reported line 101)May include surrounding context.

bash
# 删除技能目录
rm -rf ~/.openclaw/workspace/skills/credential_auditor

故障排除

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

This variant of the mismatch finding states the real capability may only be password dictionary generation while default-password matching and brute-force testing are unimplemented. Partial implementation disguised as a full credential auditor can create a false sense of coverage and cause teams to miss real authentication weaknesses.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This variant of the mismatch finding states the real capability may only be password dictionary generation while default-password matching and brute-force testing are unimplemented. Partial implementation disguised as a full credential auditor can create a false sense of coverage and cause teams to miss real authentication weaknesses.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

This variant of the mismatch finding states the real capability may only be password dictionary generation while default-password matching and brute-force testing are unimplemented. Partial implementation disguised as a full credential auditor can create a false sense of coverage and cause teams to miss real authentication weaknesses.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

This variant of the mismatch finding states the real capability may only be password dictionary generation while default-password matching and brute-force testing are unimplemented. Partial implementation disguised as a full credential auditor can create a false sense of coverage and cause teams to miss real authentication weaknesses.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

This script is explicitly designed to automate credential brute-force attempts across multiple protocols and exposes that capability directly via command-line arguments without any authorization gate, scope validation, or user warning. In the context of an agent skill named 'credential_auditor', this is especially dangerous because it operationalizes offensive password guessing against arbitrary targets and could be readily misused for unauthorized access attempts.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · INSTALL.md (reported line 20)May include surrounding context.

bash
# 创建 OpenClaw 技能目录(如果不存在)
mkdir -p ~/.openclaw/workspace/skills/

# 复制技能到 OpenClaw
cp -r . ~/.openclaw/workspace/skills/credential_auditor/

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · INSTALL.md (reported line 20)May include surrounding context.

bash
# 创建 OpenClaw 技能目录(如果不存在)
mkdir -p ~/.openclaw/workspace/skills/

# 复制技能到 OpenClaw
cp -r . ~/.openclaw/workspace/skills/credential_auditor/

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The install guide explicitly recommends Hydra, Medusa, and Ncrack, which are commonly used password-cracking and brute-force tools, but provides no warning about legal, operational, or account lockout risks. In a credential-auditing skill, normalizing installation of offensive tooling without safety boundaries increases the likelihood of misuse against unauthorized targets.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · INSTALL.md (reported line 59)May include surrounding context.

bash
# 使用 apt 安装
sudo apt-get update
sudo apt-get install hydra medusa ncrack

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · INSTALL.md (reported line 60)May include surrounding context.

bash
# 使用 apt 安装
sudo apt-get update
sudo apt-get install hydra medusa ncrack

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The uninstall section uses a destructive recursive deletion command without any cautionary language or verification step. While the path is specific, copy/paste errors or path expansion mistakes can still cause unintended data loss.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document prominently advertises brute-force credential testing and protocol attacks, but the nearby feature description does not clearly warn about concrete operational risks such as account lockouts, service disruption, alerting, or testing against unauthorized targets. In a skill designed for agent use, this omission increases the chance that users or downstream automation invoke intrusive actions without understanding likely impact on systems and data.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · PROJECT_INFO.md (reported line 109)May include surrounding context.

md
# 1. 进入项目目录

# 2. 安装到 OpenClaw
mkdir -p ~/.openclaw/workspace/skills/
cp -r . ~/.openclaw/workspace/skills/credential_auditor/

# 3. 安装 Python 依赖

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The example shows a natural-language request causing the AI to automatically invoke the credential auditing skill and return default credentials, without any gating, authorization check, or safety warning. For a security-sensitive skill, normalizing automatic execution from casual prompts materially raises misuse risk, especially when the capability includes credential discovery and brute-force testing.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 84)May include surrounding context.

  1. 安装到 OpenClaw
    bash
    # 复制技能到 OpenClaw 工作区技能目录
    mkdir -p ~/.openclaw/workspace/skills/
    cp -r . ~/.openclaw/workspace/skills/credential_auditor/
    

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
85% confidence
Finding

The skill advertises and documents capabilities that inherently involve reading reference files and writing output reports/wordlists, but it declares no explicit tool scope or permissions. In an agent environment, missing scope boundaries increases the chance the skill can access or modify files beyond what operators expect, undermining least-privilege controls.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The activation logic is broad enough to trigger on generic security-testing requests, including requests involving default passwords, password generation, and brute-force activity, without requiring clear authorization boundaries. Because this is an offensive-security-adjacent skill, broad activation materially increases misuse risk and makes accidental deployment against unauthorized targets more likely.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This Python skill uses Chinese-only natural-language strings in the module docstring, class docstring, CLI description, help text, and runtime messages. Under the policy rule, forcing a specific language without offering user choice or documenting a justified region-specific constraint is a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and user-facing descriptions are entirely in Chinese, and the command help text later in the file is also Chinese-only. This imposes a specific language on users without opt-in or any documented reason that the skill is region-specific, matching the language/locale policy violation criteria.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · README.md (reported line 111)May include surrounding context.

md
print("\n安装建议:")
        if not self.available_tools['hydra']:
            print("  - Hydra: brew install hydra (macOS)")
            print("           sudo apt-get install hydra (Ubuntu)")
        
        if not self.available_tools['medusa']:
            print("  - Medusa: brew install medusa (macOS)")

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · SKILL.md (reported line 204)May include surrounding context.

md
print("\n安装建议:")
        if not self.available_tools['hydra']:
            print("  - Hydra: brew install hydra (macOS)")
            print("           sudo apt-get install hydra (Ubuntu)")
        
        if not self.available_tools['medusa']:
            print("  - Medusa: brew install medusa (macOS)")

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · scripts/tool_integration.py (reported line 43)May include surrounding context.

python
print("\n安装建议:")
        if not self.available_tools['hydra']:
            print("  - Hydra: brew install hydra (macOS)")
            print("           sudo apt-get install hydra (Ubuntu)")
        
        if not self.available_tools['medusa']:
            print("  - Medusa: brew install medusa (macOS)")

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
70% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · scripts/tool_integration.py (reported line 47)May include surrounding context.

python
print("\n安装建议:")
        if not self.available_tools['hydra']:
            print("  - Hydra: brew install hydra (macOS)")
            print("           sudo apt-get install hydra (Ubuntu)")
        
        if not self.available_tools['medusa']:
            print("  - Medusa: brew install medusa (macOS)")