Security audit
Security checks for vulnerabilities and agentic risk
The skill is not malware, but it overstates live casino scanning and can produce misleading gambling recommendations from static, unverified data while running on a schedule.
Review before installing. Treat the rankings as a static sample calculation, not a live scan of current casino offers or a reliable financial recommendation. If installed, keep the scheduled run and BONUS_OUTPUT_FILE path under your control, and independently verify bonus terms, wagering rules, reputation, and legality before acting on any result.
The skill materially overstates its capabilities, claiming to scan 30+ casinos, calculate reputation-adjusted EV, and produce ranked results, while the analyzed behavior does not actually fetch external data or implement the advertised logic. This can mislead users into trusting fabricated or incomplete gambling recommendations, creating integrity and decision-making risk, especially where money is involved.
Without declared permissions the skill's intent is opaque and cannot be validated.
The manifest describes an active scanner over online casinos, which implies collecting current bonus data from external sources. In the implementation, all casino and bonus data are statically embedded in the BONUSES list, with no network access or scraping logic, so the behavior is materially narrower than claimed.
The skill emits rating strings "BON", "MOYEN", "FAIBLE", and "NEGATIF", which impose a French locale in user-visible output. The file does not offer language choice or explain why French output is required, which violates the language/locale policy criterion.
The code takes BONUS_OUTPUT_FILE directly from the environment and passes it to open(..., "w") without validation. In an agent or shared runtime, an attacker who can influence environment variables could cause arbitrary file overwrite within the process's permissions, potentially clobbering configuration, logs, or other sensitive files.
}
out_file = os.environ.get("BONUS_OUTPUT_FILE", "/tmp/casino_bonuses.json")
with open(out_file, "w") as f:
json.dump(output, f, indent=2)
print(f"\nCasino Bonus Hunter — {len(results)} bonuses ranked by EV (adjusted for reputation)")
The function documentation says net EV should deduct any required deposit, but the implementation sets ev_net equal to ev_gross and ignores deposit_required entirely. In a gambling decision-support skill, this can systematically overstate bonus profitability and mislead users into risking real money on offers that are materially worse than advertised.
The manifest says the skill outputs a ranked list so a user knows which bonus to hunt next, which suggests returning/displaying results. The code additionally writes JSON to an environment-controlled filesystem path, which is extra behavior not disclosed in the description.
This code writes results to a path from BONUS_OUTPUT_FILE, defaulting to /tmp/casino_bonuses.json, but there is no preceding comment, docstring warning, or confirmation indicating that the skill will create or overwrite a file. For code files, file writes should have some visible disclosure unless clearly warned elsewhere.
This file contains user-facing natural-language text in French in the module docstring and function documentation, but it does not indicate that the skill is French-only or provide any user opt-in for language preference. Under the policy, forcing a specific language without user choice is a natural-language policy concern.
No suspicious patterns detected.