Security audit
Security checks for vulnerabilities and agentic risk
This is a local stock position-size calculator with disclosed financial inputs and report generation, and I found no hidden network, credential, destructive, or system-persistence behavior.
Install only if you are comfortable running a local Python calculator and having position-sizing reports saved on disk. Treat its outputs as risk-management math, not investment advice, and confirm the inputs before using the result for a real trade.
The skill declares no allowed tool scope, yet its workflow instructs the agent to execute a local Python script via shell. That creates a permissions/behavior mismatch: an orchestrator or reviewer cannot clearly determine that shell execution is intended, which increases the risk of unintended command execution or over-privileged invocation if the skill is enabled.
The implicit triggers are broad enough that the skill could activate for general discussion about buying stocks or managing risk, even when the user did not explicitly request position sizing. In financial contexts, unintended invocation can lead to overconfident, tool-driven outputs or premature calculations based on incomplete assumptions.
The skill persists user-supplied financial calculation outputs to disk even though its stated purpose is a calculation utility. Writing reports creates an unnecessary side effect that can expose sensitive portfolio/account data, leave recoverable artifacts on shared systems, or enable storage abuse if invoked repeatedly. In this context, the behavior is not overtly malicious, but it expands the skill's data-handling surface beyond what users may reasonably expect from a position-sizing tool.
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def test_risk_pct_and_kelly_mutual_exclusive(self):
"""Both risk_pct and win_rate via CLI -> SystemExit (argparse error)."""
script = "skills/position-sizer/scripts/position_sizer.py"
result = subprocess.run(
[
sys.executable,
script,
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def test_risk_pct_and_kelly_mutual_exclusive(self):
"""Both risk_pct and win_rate via CLI -> SystemExit (argparse error)."""
script = "skills/position-sizer/scripts/position_sizer.py"
result = subprocess.run(
[
sys.executable,
script,
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def test_cli_missing_required(self):
"""Missing --account-size -> SystemExit."""
script = "skills/position-sizer/scripts/position_sizer.py"
result = subprocess.run(
[
sys.executable,
script,
No suspicious patterns detected.