agent-memory
Long-term memory runtime for AI agents — plain Markdown as the source of truth, local ranked retrieval, and an independent sleep-time Manage layer. Claude Code and Codex share one store. No API key.
AI Atlas + Technical Lab
Find agent-native AI tools by category, then verify the stack.
Browse by the job first: coding, automation, data, research, communication, and security. Then filter by MCP, Skill, CLI, Workflow, compatible agent, source trust, stars, and freshness.
01
Start from AI categories.
MCP / Skill / CLI / Workflow remain important, but they work better as filters than as the first door.
02
Newly published tools.
Cards show value first, then technical type, source evidence, and a quick path to the detail page.
corsair
Connect your users to their apps
gentle-pi
Turn Pi into el Gentleman: a senior-architect development harness with SDD/OpenSpec, subagents, strict TDD evidence, review guardrails, and skill discovery.
llm-gateway
Connect Your Agents And Harnesses With Any Provider 🦚
Easel
An open-source AI agent for social media — discover trends, create content, publish everywhere, and learn what works across Xiaohongshu, Douyin, Zhihu, Bilibili, and more.🎨一个开源的 AI 社交媒体智能体——发现热点趋势、创作内容、一键发布至各大平台，并学习分析哪些内容真正有效，覆盖小红书、抖音、知乎、哔哩哔哩等平台。
03
Agent compatibility comes next.
Use agent pages after picking the problem space. They explain which tools, skills, MCPs, and workflows fit each agent.
global Claude Code
Anthropic's terminal-native coding agent for large repositories, MCP-connected tools, skills, hooks, and team workflows.
china CodeGeeXA China-origin IDE coding assistant focused on completion, explanation, and editor-side developer assistance.
global CodexOpenAI's coding agent for repository implementation, review, verification, and tool-assisted engineering work.
global CursorAn editor-first AI coding environment with agentic IDE workflows, rules, memories, MCP, and cursor-agent CLI support.
global Hermes AgentNous Research's open-source, terminal-oriented agent with skills, MCP integration, local execution, and a large Skills Hub ecosystem.
china Kimi CLI / Kimi AgentMoonshot AI's Kimi agent surface for long-context research, document-heavy workflows, and Kimi Code CLI development tasks.
04
Signals and Learn support the directory.
They explain why the ecosystem is moving, without taking over the tool-first homepage.
open_license
Hybrid LLM-RL Red Teaming Framework Exposes AI Security Gaps
This paper introduces an autonomous red teaming framework that combines large language models with reinforcement learning to generate adaptive, multi-stage attack campaigns against AI-enabled security systems. Testing in high-fidelity enterprise simulations reveals that standalone LLM agents cannot sustain complex attacks, while hybrid LLM-RL approaches achieve significantly higher compromise rates, exposing critical vulnerabilities in current AI security defenses.
open_licenseAdaptive Multi-Agent Framework for Workflow Automation
This paper introduces a multimodal multi-agent framework that automatically executes complex workflows by first constructing a topological knowledge graph from fragmented execution logs (offline), then using adaptive Retrieval-Augmented Generation (RAG) over that graph during inference. A closed-loop verification protocol enables agents to self-correct and navigate non-stationary scenarios, outperforming linear approaches in reliability and semantic awareness.
learnRead the field guides
Short explanations for MCP, Skills, Workflows, and agent selection.