Security audit
Security checks across malware telemetry and agentic risk
Input Guard is a coherent defensive scanner, but it needs Review because optional LLM scanning can use locally configured OpenClaw gateway API keys and send scanned text to external providers without clear top-level disclosure.
Install only if you are comfortable with a defensive scanner that has optional networked modes. Use the default pattern-only mode for sensitive private content. Avoid --llm, --llm-only, or --llm-auto unless you accept sending scanned text to the selected LLM provider and potentially using API keys already configured in OpenClaw gateway. Enable channel alerts and MoltThreats reporting only for destinations where sharing source URLs and detection details is acceptable.
def _call_openai(api_key: str, model: str, system_prompt: str, user_text: str,
timeout: int = 30) -> Dict:
"""Call OpenAI Chat Completions API."""
response = requests.post(
"https://api.openai.com/v1/chat/completions",
headers={
"Authorization": f"Bearer {api_key}",def _call_anthropic(api_key: str, model: str, system_prompt: str, user_text: str,
timeout: int = 30) -> Dict:
"""Call Anthropic Messages API."""
response = requests.post(
"https://api.anthropic.com/v1/messages",
headers={
"x-api-key": api_key,def _call_openai(api_key: str, model: str, system_prompt: str, user_text: str,
timeout: int = 30) -> Dict:
"""Call OpenAI Chat Completions API."""
response = requests.post(
"https://api.openai.com/v1/chat/completions",
headers={
"Authorization": f"Bearer {api_key}",def _call_anthropic(api_key: str, model: str, system_prompt: str, user_text: str,
timeout: int = 30) -> Dict:
"""Call Anthropic Messages API."""
response = requests.post(
"https://api.anthropic.com/v1/messages",
headers={
"x-api-key": api_key,timeout: int = 30) -> Dict:
"""Call OpenAI Chat Completions API."""
response = requests.post(
"https://api.openai.com/v1/chat/completions",
headers={
"Authorization": f"Bearer {api_key}",
"Content-Type": "application/json",66/66 vendors flagged this skill as clean.
Detected: suspicious.prompt_injection_instructions