Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed, user-driven Perchance image generator, with the main consideration being that prompts are sent to a third-party website.
Install only if you are comfortable sending image prompts to Perchance.org. Do not enter secrets, private business information, regulated data, or personal details in prompts unless you have reviewed and accepted that third-party processing risk.
The skill explicitly sends user-supplied prompts to Perchance.org, an external third-party service, but the description and usage text do not clearly warn users that their input leaves the local environment. This can expose sensitive prompts or embedded personal/business data to an external service without informed user consent, making the omission a real privacy and transparency issue.
This duplicate finding points to the same outbound transmission behavior: prompt content is sent to perchance.org over the network. Within a skill ecosystem, such hidden external transmission increases risk because users may unintentionally disclose secrets, personal data, or internal project information.
# Send to Perchance
try:
response = requests.post('https://perchance.org/new-image-gen-by-rs118',
json={'prompt': prompt, 'orientation': orientation})
if response.status_code == 200:
This duplicate finding points to the same outbound transmission behavior: prompt content is sent to perchance.org over the network. Within a skill ecosystem, such hidden external transmission increases risk because users may unintentionally disclose secrets, personal data, or internal project information.
# Send to Perchance
try:
response = requests.post('https://perchance.org/new-image-gen-by-rs118',
json={'prompt': prompt, 'orientation': orientation})
if response.status_code == 200:
User-supplied prompt text is sent to a third-party service without any explicit disclosure, consent flow, or privacy notice. In a skill context, users may enter sensitive or proprietary text, and silent transmission to an external endpoint can expose confidential data to an untrusted processor.
The docstring at L182 describes openclaw_run as opening a terminal-based web browser, which implies a terminal or text-mode interface. In reality, the function creates a QApplication, constructs Qt widgets, shows GUI windows, and runs app.exec_(), so the documented intent directly contradicts the implemented behavior.
No suspicious patterns detected.