Search

AutoGen integration for the Chio protocol -- capability-scoped function registration, GroupChat role scoping, nested chat attenuation

Official Python SDK for AgentShield — prompt injection detection for LLM applications.

Local inventory of the AI agent surface on your machine. No account, no network, no upload.

AI skill: authentication and authorization conventions for web applications.

AI Threat Detection Protocol — reference SDK for detecting prompt injection, jailbreaks, data exfiltration and tool abuse in LLM apps and agents.

The authorization layer MCP doesn't have. JIT proxy for AI agents.

Filesystem sandbox toolset for PydanticAI agents with LLM-friendly errors

Wire Protocol CLI — init, validate, run, registry

A spend firewall for AI agents making x402 payments. Enforces caps, allowlists, signature-integrity binding, and a kill switch before any payment is signed. Pre-alpha; v1 is anti-drain + signature integrity, EVM-only.

APort Agent Guardrail — shared core for AI agent and LLM frameworks (pre-action authorization)

Framework-independent, fail-closed human approval middleware for consequential AI agent tool calls.

Run the OpenAI Agents SDK's sandbox commands in local Hyperlight micro-VMs

LLM-powered OSINT agent that discovers validator infrastructure beyond on-chain data. Sui, Solana, and more.

A governance-first multi-agent framework that treats agents as an organization, not just a router.

AWS SDK for JavaScript Sagemaker Client for Node.js, Browser and React Native

Read-only static scanner for AI-agent codebases: finds dangerous action sinks, proves reachability, attributes guard coverage, and rates findings honestly (reachable-verified vs install-liability).

DuckDB multi-org agent swarm with Spec Engine, 175+ SQL macros, meta-learning, MCP Apps, and smart extensions

A local proxy that transparently injects API credentials into outbound HTTPS requests. Your apps and AI agents never hold real secrets.

Framework-agnostic core for AI Kit - streaming, agents, state management, and LLM primitives

Tiered permission gate for omp (oh-my-pi): local LANCET-Nano triage plus a model adjudicator, inspired by Claude Code auto mode