Mfa Servicenow MCP Server

In the server's implementation (`src/servicenow_mcp/auth/_http_session.py:297`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: rust_env = False # Skip .netrc / env proxy cookies # urllib3 retries are deliberately disabled (connect=0, read=0).

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/auth/_http_session.py

In the server's implementation (`src/servicenow_mcp/browser/evaluate.py:8`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: . It is compiled with ``new Function('return (' + src + ')')``, so a statement body is a SyntaxError rather than a silen

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/browser/evaluate.py

In the server's implementation (`src/servicenow_mcp/tools/audit_tools.py:73`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: , "eval_usage", "Direct eval() — code injection risk"), (r"\.innerHTML\s*=", "xss_risk", "innerHTML assignment —

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/tools/audit_tools.py

In the server's implementation (`src/servicenow_mcp/auth/auth_manager.py:634`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: result = subprocess.run( [sys.executable, "-m", "playwright", "install", "chromium"],

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/auth/auth_manager.py

In the server's implementation (`src/servicenow_mcp/browser/window.py:834`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: file_dir) process = subprocess.Popen( # noqa: S603 - executable path comes from Playwright [executable, *ar

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/browser/window.py

In the server's implementation (`src/servicenow_mcp/setup_installer.py:792`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ) try: subprocess.run(cmd, check=True, timeout=600) return "installed" except FileNotFoundE

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/setup_installer.py

In the server's implementation (`src/servicenow_mcp/tools/repo_tools.py:51`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: -> str: completed = subprocess.run( ["git", "-C", str(repo_path), *args], check=False, captu

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server src/servicenow_mcp/tools/repo_tools.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_audit_tools.py:94`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: code = "var x = eval('alert(1)');" patterns = _compile_risk_patterns() findings = _scan_code(cod

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_audit_tools.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_debug_window.py:2475`): Evaluating strings as code is the most direct RCE primitive; if any tool input reaches it, the server executes attacker-chosen code. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: s.length") assert "new Function('return (' + src + ')')" in script # The source is embedded as a JSON string li

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_debug_window.py

In a packaging/dev/install script (shipped, but not the server runtime) (`scripts/build_desktop_release.py:50`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: n(cmd), flush=True) subprocess.run(cmd, cwd=ROOT, env=env, check=True) def _zip_dir(source_dir: Path, zip_path: Pa

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server scripts/build_desktop_release.py

In a packaging/dev/install script (shipped, but not the server runtime) (`scripts/check_real_identities.py:231`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: try: out = subprocess.run( ["git", "diff", "--cached", "--name-only", "--diff-filter=ACMR"],

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server scripts/check_real_identities.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_cli.py:833`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: t sys result = subprocess.run( [sys.executable, "-m", "servicenow_mcp", "--version"], c

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_cli.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_json_fast.py:71`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ("OK") """ result = subprocess.run( [sys.executable, "-c", code], capture_output=True, text=

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_json_fast.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_json_fast_fallback.py:37`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ode: str): result = subprocess.run( [sys.executable, "-c", code], capture_output=True, text=

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_json_fast_fallback.py

In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_auth_manager_browser.py:817`): Loading a module chosen at runtime (from a variable) can pull in and run attacker-influenced code paths. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.

Evidence: ASIC, basic=__import__( "servicenow_mcp.utils.config", fromlist=["BasicAuthConfig"]

Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.

Location: server tests/test_auth_manager_browser.py