What Tablif read on Bedrock Guardrail Firewall's site
Bedrock Guardrail Firewall describes itself as: “Bedrock Guardrail Firewall”
- Stage: Live, no waitlist
What it is
Calls itself “policy enforcement layer”.
- Calls itself: policy enforcement layer
“Bedrock Guardrail Firewall is an open-source, privacy-first policy enforcement layer for generative AI.”
Where it puts itself
Rules out “credentials”.
- Says it will not: credentials
“It excludes blocked content, credentials, local paths, queue URLs, bucket names, and keys.”
What it costs
No free plan.
- Free plan: no
“Apache-2.0 license”
Who it sells to
Sells to developers.
- Sells to: developers
“Bedrock Guardrail Firewall is an open-source, privacy-first policy enforcement layer for generative AI.”
How it runs
Runs on Linux and Windows. Runs locally and works offline. Distributed through the PyPI. Offers an API. Open source. AI is the core of the product.
- Runs on: Linux · Windows
- Runs where: on the device, offline
- Distributed via: PyPI
- API: yes
- Open source: yes
- AI role: core to the product
“The core supports Windows and Linux with Python 3.12 through 3.14.”
“Offline by default, fail-closed in production, and backed by metadata-only, tamper-evident audit trails.”
What it works with
Integrates with Amazon Bedrock and Microsoft Presidio. Names GitHub on its pages.
- Integrates with: Amazon Bedrock · Microsoft Presidio
- Names: GitHub
“Amazon Bedrock Guardrails or Microsoft Presidio.”
“Bedrock Guardrail Firewall is an open-source, privacy-first policy enforcement layer for generative AI.”
What it offers as proof
Live, no waitlist. Claims “218 automated tests”. Publishes docs.
- Number claimed: 218 automated tests
“You must be signed in to star a repository”
“218 automated tests across the core, AWS SDK contract, Presidio integration, release validation, and sanitized demo”
Sources
Tablif read these pages, last on 2 Nov 2026. Everything above is what the product says about itself; Tablif does not verify every claim.
Similar products in Tablif Town
- Vibe Guardrails: Drop-in rules that keep AI coding tools from leaking keys
- Muzzle: the firewall for your AI coding agents
- Secure AI - Privacy-first AI: AI you can trust — for your data and your company's data
- CacheGuard: Turn any machine into an internet security gateway
- masked.chat: A local privacy shield for every AI interaction
- Claustra: Security for AI coding agents
- SafeLabs Guard: An independent verification layer for multimodal AI agents.
- EraseAI Firewall: Stop secrets and PII before you hit Send in Public LLMs