Search
GitHub Actions
A TypeScript framework for agents, typed tools and durable workflows.
nelsoncc
Runtime governance for tool-using AI agents in Java — budget enforcement, tool authorization, human approval, and loop brakes.
...morepierlink
麦麦护栏:把麦当劳账号交给 AI 之前,先装个刹车。本地 MCP 安全层:下单等写操作需本机确认、消费上限、只读模式、审计日志,Token 只在本机。Human-in-the-loop guard proxy for McDonald's China MCP.
...morendpvt-web
Let your cloud AI agent work on your Mac - safely. Risky actions wait for your yes, and what you approve is exactly what runs.
...morepaepckehh
agentic assisted automatic git commit, sign, tag, push with dedicated roles for agent/human, generate with one command a detaild commit message analysis, commits, signs, tags automaticly, leaves the review & git push to human only
...morevantaverse
🤖 The world's first truly agentic code review tool. AI that thinks, explores & analyzes your code autonomously. 70-85% cost savings vs traditional tools. Built with Vercel AI SDK v6.
...moreliuyanghejerry
CI-native agent CLI tool for deterministic pipeline gating.
esammy
Runtime observability, policy enforcement, and audit trails for AI agents
npmtorrixai
TypeScript SDK for Torrix, self-hosted observability for AI agents: record OpenAI, Anthropic and Vercel AI SDK calls with zero dependencies
...morederivativelabs
TypeScript SDK for AI agents to access LivePort tunnels
honestDS
MonoLight 是一个专注于安全执行与人机协同的通用自主智能体(General Autonomous Agent)运行时。
LoopGrid contributors
Native Pydantic AI capability for signed, verifiable consequential agent action evidence
mgillr
Poison-resistant, Byzantine-fault-tolerant shared chat memory for LangChain agent swarms — one prompt-injected agent can't corrupt the shared history.
...moreA file-based framework for asking clear research questions, checking sources, and recording what the evidence supports.
GitHub Actions
Dual-layer compliance classifier for AI agent communications
Canyon-Li
透明可审计的本地智能体终端:对开源 agent 框架做安全审计与加固(RCE/沙盒绕过/注入),并以 50 条注入基准 × 37 条能力基准双维度量安全边界
Major Labs
Per-task budget, loop detection, and kill-switch middleware for agent LLM calls. v0, deterministic and dependency-free.
Tom Farley
ScopeBlind evidence wrapper for CrewAI tool calls. Automatically creates Veritas Acta receipts for every tool invocation and delegation.
...more