Personal data is masked before it reaches the model and restored in the response.
Prompt injection, jailbreaks and unsafe tool calls are stopped before they do harm.
Self-hosted, with audit trails and reports for the EU AI Act, GDPR and KVKK.
One line of code
Wrap your Claude client with guard() and every request passes through 15+ guards before it reaches the model. Works the same way with OpenAI and other providers.
from anthropic import Anthropic
from archeguard import guard
client = guard(Anthropic())
response = client.messages.create(
model="claude-sonnet-5-5",
max_tokens=1024,
messages=[{"role": "user", "content": prompt}],
)Compliance
Tamper-evident audit trails and evidence reports, generated from your own deployment.