naptic-guard
AI data governance, privacy enforcement and information security
Downloadable command-line software that audits the Amazon Bedrock Guardrails protecting your AI agents. It reports exactly what is missing: PII policies that were never configured, prompt-injection filters that were never enabled, and credentials that are anonymized when they should be blocked. Every call it makes is read-only.
Install
pip install git+https://github.com/NapticStacks/naptic-guard@v0.1.0Run
naptic-guard audit --profile my-profile --region us-east-1What it checks
- NG001A sensitiveInformationPolicy exists, so PII can be anonymized or blocked.
- NG002A PROMPT_ATTACK content filter is enabled against prompt injection and jailbreaks.
- NG003Credentials and financial identifiers are BLOCK, not ANONYMIZE.
