70强项目观察之新智基座：Agent正在从“会聊”走向“会干活”
摘要
InfoQ 分析 GOAI「新智基座 Agent Infra」赛道 15 个决赛项目，认为竞争重点正从模型能力转向系统能力，权限隔离、证据链、Human Gate、状态回读与回滚成为共同工程问题。文章以 DataFlow-Agent、CodeNotary、CyberGuard、RepoMesh 等项目为例，指出 Agent 正从生成答案走向交付结果，多 Agent 从角色扮演走向职责协作。
相关推荐
morluto/rea
Reverse engineer anything with agents, from app behavior down to native binaries.
Cursor 转 Codex 大半个月，聊聊我的真实感受
Cursor 让我第一次强烈感受到 AI 编程的震撼，Codex 则让我第一次真正感觉到，AI 不只是编程助手，而是一个真正的超级Agent，Codex 明显更接近下一阶段的形态。
vectorize-io/hindsight
Hindsight: Agent Memory That Learns
alibaba/open-code-review
Secure, fast, efficient, battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in multi-language ruleset (NPE, thread-safety, XSS, SQL injection), OpenAI & Anthropic compatible.
affaan-m/ECC
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.