用命令行驱动浏览器的自学习 Agent，基于 Playwright 自动操作网页，并在使用中积累经验、越用越顺手，适合做网页自动化与数据采集。
用 Rust 写的高性能向量图神经网络数据库，把向量检索、GNN 推理和智能体记忆放在一起，主打实时决策和低延迟自学习。
让 AI 代理直接操控 macOS 桌面完成各类操作，靠系统辅助功能接口而非截图识别界面，还能自我学习并复用操作流程。
更新内容 Routine rebuild, no agent-specific changes.
<!-- agentconnect-release-summary:start -- Summary - Single Sign-On — Self-hosted deployments can sign in through their own identity provider over OIDC or SAML, without c
<!-- agentconnect-release-summary:start -- Summary - Auto Session Mode — A new Auto mode runs a conversation through a coordinator that reads incoming messages and hands
🎉 Welcome to the 1.7.0-alpha.1 release of the agentgateway project! Agentgateway v1.7.0 is expected next week. Artifacts Docker images are available: cr.agentgateway.dev
…/giskard-oss/compare/giskard-llm/v1.0.3...giskard-agents/v1.0.5
…Container images（amd64 + arm64 · 匿名可拉）: - ghcr.io/agentpit-io/hunter-community-api:v2.1.0 - ghcr.io/agentpit-io/hunter-community-web:v2.1.0 - ghcr.io/agentpit-io/hunter-c