Back to skill

Security audit

Ai Agent Tools

Security checks for vulnerabilities and agentic risk

Overview

The utility code is mostly straightforward, but the installation guidance tells users to fetch mutable GitHub code and includes unnecessary sudo execution, so it needs review before use.

Install only from the reviewed local artifact or a pinned, verified release. Do not run this library with sudo or administrator privileges, and only expose its file read/write tools to an agent with explicit path limits and user-approved output locations.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:19
Finding

Unverified Remote Python Payload Retrieved from a Mutable Branch

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
INSTALLATION.md:262
Finding

Unnecessary Root Execution Recommended as a Permission Workaround

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (11)

YARA rule 'agent_skill_remote_bootstrap_execution': Remote script or code download followed by execution/bootstrap installation [agent_skills]

High
Category
YARA Match
Confidence
90% confidence
Finding

The installation guide instructs users to install directly from a live GitHub repository and also to download raw code for execution, which bypasses stronger supply-chain controls such as signed releases, pinned commits, hashes, or vetted package publication. In a skill/install context this is more dangerous because the document normalizes immediate trust and execution of remote code, so repository compromise or upstream tampering could lead to arbitrary code execution on user systems.

Content

Scanner excerpt · INSTALLATION.md (reported line 45)May include surrounding context.

Method 2: Clone from GitHub (Recommended for Development)

bash
# Clone the repository
git clone https://github.com/cerbug45/ai-agent-tools.git

# Navigate to the directory
cd ai-agent-tools

# Test the installation
python ai_agent_tools.py

Method 3: Install as Python Package (Coming Soon)

bash
# Using pip (when published to PyPI)
pip install ai-agent-tools

# From GitHub directly
pip install git+https://github.com/cerbug45/ai-agent-tools.git

Method 4: Add as Git Submodule (For Larger Projects)

bash
# Add to your project as a submodule
git submodule add https://github.com/cerbug45/ai-agent-tools.git libs/ai-agent-tools

# Update the submodule
git submodule update --init --recursive

🐍 Python Version Requirements

  • Minimum: Python 3.7
  • Recommended: Python 3.9 or higher
  • Tested on: Python 3.7, 3.8, 3.9, 3.10, 3.11

Check your Python version:

bash
python --version
# or
python3 --version

📋 System Requirements

Operati

Memory Manipulation

High
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill manipulates agent memory, state, or stored context. Memory corruption can alter personality, override safety rules, or cause unpredictable behavior.

Content

Scanner excerpt · SKILL.md (reported line 304)May include surrounding context.

text

### 2. Memory Management
Clear memory when no longer needed:

```python
memory = MemoryTools()

Rp1

Medium
Category
MCP Rug Pull
Confidence
75% confidence
Finding

Docker image references without a specific tag (:latest is implicit) or digest (@sha256:...) can be silently replaced by a malicious image.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
84% confidence
Finding

The guide recommends running sudo python ai_agent_tools.py as a troubleshooting step, which encourages executing downloaded code with elevated privileges. In the context of this installation document, users are also told to fetch code directly from GitHub, so combining remote code retrieval with root execution materially increases the chance of full-system compromise if the file is malicious or tampered with.

Content

Scanner excerpt · INSTALLATION.md (reported line 268)May include surrounding context.

chmod +x ai_agent_tools.py

Or run with appropriate permissions

sudo python ai_agent_tools.py

text

### Issue: Python Version

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

This markdown file documents write and file-management capabilities such as writing files, saving outputs, and later clearing memory, but it does not include any warning that these actions can overwrite data or remove stored state. For markdown files, SQP-2 applies when descriptions omit warnings about behaviors that could affect user data or system integrity.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The documentation specifies phone validation as 'Turkish format' and uses Turkish-format examples elsewhere, which imposes a locale-specific assumption in a general-purpose utility library. Under SQP-3, locale constraints should either be optional for the user or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The file's human-facing docstrings, status messages, and example output are entirely in Turkish, which effectively fixes the skill's natural-language interface to a single language. There is no indication that users can choose another language or that the Turkish-only behavior is a justified regional constraint.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code file defines a general-purpose write operation that creates directories and overwrites file contents, but provides no confirmation prompt or explicit user disclosure before the write occurs. Although there is a success return string after completion, it does not warn the user in advance, and the helper is broad enough that file modification may not be obvious from the calling context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The JSON save function persists arbitrary data to disk and creates parent directories, but the code only reports success after the action is complete. There is no confirmation mechanism, pre-write disclosure, or explicit warning in the surrounding documentation about saving or overwriting user data.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document shows the expected test output entirely in Turkish, which implies a fixed language experience for users following the guide. Because no user opt-in, alternative language option, or justification for the locale is provided, this appears to violate the language/locale policy criteria.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

This markdown file describes file operations including writing files, and later shows save_json(..., "output.json") and save_json(data, "users.json") as normal usage. There is no accompanying warning about overwriting existing files, choosing safe paths, or reviewing outputs before saving, which is a user-data integrity concern for a skill aimed at AI agents.

Content

No source excerpt is available for this finding.