Back to skill

Security audit

Highgo Db

Security checks for vulnerabilities and agentic risk

Overview

This skill is a database helper, but it needs Review because it handles real database credentials unsafely and can load arbitrary local driver code.

Install only if you are comfortable reviewing a database tool that can execute write-capable SQL. Use a least-privileged, preferably non-production HighGo account, avoid putting passwords in command lines, and do not use --driver except with a trusted, integrity-verified local driver directory.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/execute_query.py:7
Finding

Caller-Controlled Driver Path Enables Arbitrary Python Code Execution

Content
View full analysis

Vulnerability Details

File Location: scripts/execute_query.py:7-35
Vulnerability Type: Unsafe dynamic module loading
Risk Level: High

Vulnerable Code

python
def execute_query(dsn, sql, driver_path=None):
    # If no driver_path provided, look in the skill's assets/driver/psycopg2
    if driver_path is None:
        script_dir = os.path.dirname(os.path.abspath(__file__))
        driver_path = os.path.join(script_dir, "..", "assets", "driver", "psycopg2")
    
    if not os.path.exists(driver_path):
        return {"status": "error", "message": "Driver not found at {}".format(driver_path)}

    # Manual temporary directory for Python 2.7 compatibility
    tmp_dir = tempfile.mkdtemp()
    try:
        target_path = os.path.join(tmp_dir, "psycopg2")
        if not os.path.exists(target_path):
            os.symlink(os.path.abspath(driver_path), target_path)
    
        # Add the driver path to LD_LIBRARY_PATH for the .so libraries
        env = os.environ.copy()
        driver_abs_path = os.path.abspath(driver_path)
        if "LD_LIBRARY_PATH" in env:
            os.environ["LD_LIBRARY_PATH"] = "{}:{}".format(
                driver_abs_path, env['LD_LIBRARY_PATH'])
        else:
            os.environ["LD_LIBRARY_PATH"] = driver_abs_path
    
        # Add the temp directory to sys.path
        sys.path.insert(0, tmp_dir)
    
        try:
            import psycopg2
            from psycopg2 import extras

The externally selectable path is exposed at scripts/execute_query.py:66:

python
parser.add_argument("--driver", required=False,
                    help="Path to psycopg2 directory (optional, uses built-in if omitted)")

Technical Analysis

The --driver argument accepts any existing filesystem path. The script creates a symbolic link from that path to a temporary package named psycopg2, places the temporary directory at the beg ...[truncated 2121 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the --driver option if external driver substitution is not strictly required.
  2. Package and import one pinned, reviewed driver from a fixed Skill-owned directory.
  3. If external drivers are required, allow only paths beneath an administrator-configured trusted root.
  4. Resolve paths with realpath() and reject paths, parent components, or files containing symbolic links.
  5. Reject driver directories writable by untrusted users and validate file ownership and permissions.
  6. Verify every driver file against a signed manifest or pinned cryptographic hashes before importing it.
  7. Do not add caller-controlled directories to LD_LIBRARY_PATH; use fixed, integrity-verified native-library locations.
  8. Execute unavoidable third-party drivers in an isolated, least-privileged process with restricted filesystem and network access.
  9. Avoid retaining untrusted entries in global sys.path or process environment state.

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/execute_query.py:64
Finding

Database Passwords Are Exposed Through Command-Line DSN Arguments

Content
View full analysis

Vulnerability Details

File Location: scripts/execute_query.py:64-72; documented usage at SKILL.md:33-35 and README.md:46-48
Vulnerability Type: Sensitive information exposure through process arguments
Risk Level: Medium

Vulnerable Code

python
parser = argparse.ArgumentParser(
    description="Execute SQL on HighGo DB using built-in psycopg2")
parser.add_argument("--dsn", required=True, help="Connection DSN")
parser.add_argument("--sql", required=True, help="SQL query or path to SQL file")
parser.add_argument("--driver", required=False,
                    help="Path to psycopg2 directory (optional, uses built-in if omitted)")

args = parser.parse_args()

sql_content = args.sql
if os.path.exists(args.sql):
    with open(args.sql, "r") as f:
        sql_content = f.read()
            
result = execute_query(args.dsn, sql_content, args.driver)

SKILL.md:33-35 instructs users to include a password directly in the command:

bash
python2 highgo-db/scripts/execute_query.py \
  --dsn "host=10.238.18.128 port=5866 dbname=ficc user=fic password='PASSWORD' options='-c search_path=system'" \
  --sql "SELECT count(*) FROM sys_user;"

README.md:46-48 documents the same pattern:

bash
python2 highgo-db/scripts/execute_query.py \
  --dsn "host=[IP] port=[PORT] dbname=[DB] user=[USER] password='[PASS]' options='-c search_path=system'" \
  --sql "SELECT count(*) FROM sys_user;"

Technical Analysis

The prescribed interface embeds the database password in a command-line argument. Command-line arguments can be exposed through shell history, process inspection facilities, terminal-session recording, CI/CD logs, Agent execution telemetry, debugging output, and monitoring systems.

Shell quoting controls how the shell parses the argument but does not prevent the resulting DSN from appearing in the target process's argument vector. The script provides no prote ...[truncated 1342 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove passwords from command-line DSN examples and explicitly warn users not to place secrets in process arguments.
  2. Support a permission-restricted PostgreSQL password file or connection-service file.
  3. Alternatively, read the password from protected standard input using a non-echoing prompt.
  4. Integrate with an approved secret manager for automated environments.
  5. Allow a DSN without a password and obtain the secret separately at runtime.
  6. Ensure errors, logs, telemetry, and returned JSON redact passwords and complete credential-bearing DSNs.
  7. Apply restrictive permissions to all credential files and reject files that are accessible to other users.
  8. Rotate any credentials previously entered through logged command lines.
  9. Use a dedicated least-privileged database account so credential compromise has limited impact.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (15)

Env Variable Harvesting

High
Category
Data Exfiltration
Confidence
60% confidence
Finding

Code enumerates, copies, or searches environment variables for secrets. Bulk environment access can collect credentials unrelated to the skill's stated purpose.

Content

Scanner excerpt · scripts/execute_query.py (reported line 25)May include surrounding context.

python
os.symlink(os.path.abspath(driver_path), target_path)
        
        # Add the driver path to LD_LIBRARY_PATH for the .so libraries
        env = os.environ.copy()
        driver_abs_path = os.path.abspath(driver_path)
        if "LD_LIBRARY_PATH" in env:
            os.environ["LD_LIBRARY_PATH"] = "{}:{}".format(driver_abs_path, env['LD_LIBRARY_PATH'])

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The README includes an example command that passes a full DSN with username and password directly on the command line and runs a live SQL query against a real database. This is risky because shell history, process listings, logs, and screenshots can expose credentials, and users may copy the example into production-like environments without understanding the exposure.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
82% confidence
Finding

The skill advertises code-like capabilities that can access environment data and read files, but it does not declare any explicit tool scope or permission boundaries. This creates ambiguity about what the skill is allowed to access, increasing the risk of unintended secret exposure or local file disclosure when the skill is invoked in an agent environment.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation includes an example DSN with inline username and password, normalizing a practice that can expose credentials through shell history, process listings, logs, and copied transcripts. In this skill's context, the risk is heightened because it is specifically designed for direct database access, so leaked credentials could immediately enable unauthorized access to a HighGo database.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The logging connection/cursor classes log every executed query by default, and self.query may include fully rendered SQL text with parameter values. This can expose secrets and sensitive data such as credentials, tokens, personal data, or business data to files or application logs, especially because the default filter() performs no redaction.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

Although the skill description emphasizes a built-in HighGo psycopg2 driver, the code accepts an arbitrary external driver path and then imports code from that path. In a skill context, this expands trust from bundled assets to attacker-controlled local content, enabling arbitrary code execution through a malicious Python package or native library loaded as psycopg2.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest frames the skill as providing a custom psycopg2-based connection to HighGo DB, which suggests connectivity support rather than unrestricted database mutation. The code accepts arbitrary SQL, allows multiple statements, and always commits, enabling INSERT/UPDATE/DELETE/DDL behavior beyond a simple connection utility.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This code executes arbitrary SQL provided via CLI and then commits the transaction, which can modify or destroy database data. Although the script name and argument names imply query execution, there is no confirmation prompt, warning print, or explanatory comment/docstring disclosing that supplied SQL will be executed and committed against the target database.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The file presents all operational instructions in a single forced language and does not provide an alternative language option, opt-in, or justification for the locale restriction. Under the stated policy, forcing a specific language without user choice is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · assets/driver/psycopg2/_range.py (reported line 172)May include surrounding context.

python
if not isinstance(other, Range):
            return NotImplemented
        for attr in ('_lower', '_upper', '_bounds'):
            self_value = getattr(self, attr)
            other_value = getattr(other, attr)
            if self_value == other_value:
                pass

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · assets/driver/psycopg2/_range.py (reported line 173)May include surrounding context.

python
return NotImplemented
        for attr in ('_lower', '_upper', '_bounds'):
            self_value = getattr(self, attr)
            other_value = getattr(other, attr)
            if self_value == other_value:
                pass
            elif self_value is None:

Dynamic attribute access via getattr()

Low
Category
Dangerous Code Execution
Confidence
50% confidence
Finding

Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.

Content

Scanner excerpt · assets/driver/psycopg2/_range.py (reported line 203)May include surrounding context.

python
return self.__gt__(other)

    def __getstate__(self):
        return {slot: getattr(self, slot)
            for slot in self.__slots__ if hasattr(self, slot)}

    def __setstate__(self, state):

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
96% confidence
Finding

This markdown file presents all user-facing instructions in Chinese and does not indicate that the content is region-specific or provide any language/locale opt-in. Under the policy rule for natural-language violations, forcing a specific language without user choice is a reportable issue.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The script accepts a DSN from the command line and uses it to connect to the database, which may include embedded credentials or other sensitive connection details. There is no warning in code comments or CLI help that sensitive credentials may be supplied and used by this script.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

If --sql points to an existing path, the program opens and reads that file, turning the skill into a local file reader in addition to a database connectivity helper. Local file ingestion is not mentioned in the manifest and is not necessary to the narrowly stated purpose of supplying a built-in database driver.

Content

No source excerpt is available for this finding.