Coinbase wallet infrastructure for AI agents: TEE-protected keys, spend limits, KYT screening, and x402 payments.

Mistral Shieldstral 1.0: live compact multimodal moderation model. Apache 2.0.

Visa agent credentials. Dashboard fees are not the agentic SKU. Page still says ~120ms.

Agent payments stack: accounts, cards, processing, billing, identity, observability, MCP.

Previous Llama Guard snapshot. Not the 2026 safety default.

Write comprehensive clinical reports - case reports (CARE guidelines), diagnostic reports (radiology/pathology/lab), clinical trial reports (ICH-E3, SAE, CSR.

A PreToolUse hook that blocks destructive Git and file commands and secret access before AI coding agents run them, defeating reordered flags, shell wrappers, and one-liners.

Microsoft Execution Containers: an open, cross-platform sandboxed code-execution system for running untrusted model output, plugins, and agent tools.

An open-source sandbox engine that applies OS isolation and policy rules to AI agent tools, network access, and MCP calls.

Amazon Bedrock is AWS's platform for building generative AI apps and agents, with a choice of foundation models, Guardrails and in-region inference.

Anthropic's enterprise code security scanning, now powered by Claude Mythos 5 in public beta, tracing data across files to report CWE categories, severity, and suggested fixes.

NVIDIA's open-source LLM vulnerability scanner with dozens of plugins and thousands of prompts for probing jailbreaks, prompt injection, and data leakage.

NVIDIA's open-source toolkit for adding programmable guardrails to LLM conversational applications, with Colang flows for input, dialog, retrieval, and output rails.

Decentralized L1 for verifiable AI: ZKML, TEE, Vanilla, HACA, PIPE, x402, Model Hub. No invented dollar table.

Open-source AI penetration testing tool. Autonomous AI agents that run your app, find vulnerabilities, and validate them with working PoCs. Apache-2.0.

Anthropic's most capable model for cybersecurity and biology, gated behind invite-only trusted access. Same engine as Claude Fable 5.1 with relaxed safeguards for vetted organizations.

Google's new frontier model for long-horizon coding, enterprise knowledge work, and cyber defense, with a 1M output-token limit and phased Fairwind access.

Claude Code skill wrapping Varlock so secrets never reach the agent's context: validate env vars with masked output, inject at runtime, read only the schema.