这是个本地化的智能体技能栈管理工具,叫AAS Core,当前版本V15.9.0。它让Codex或Claude检查你的项目后,从本地完整目录里挑选精确的技能,但不会替你排序或推荐,只做只读验证。选好的技能可以存成aas stack.json文件,生成不可变计划再动手改代码。核心功能是目录搜索、技能选择、清单验证和规划诊断,适合需要严格把控AI编码流程、确保每一步都可复现可审查的团队。
查看英文原文
AAS Core — Agentic Awesome Skills Local, agent owned skill stacks for coding agents—from complete catalog access to a reproducible, reviewable plan. Current release: V15.9.0. This release includes AAS Core for complete local catalog search, agent owned selection, manifest validation, planning, and diagnosis. Apply and recovery remain experimental and outside the supported preview path. Codex or Claude inspects your project and chooses exact skills from the complete local AAS catalog. AAS Core does not rank or recommend them: its read only compose stack tool validates the agent owned selection in memory, and a client or the aas CLI can persist it as aas stack.json and produce an immutable plan before any target change. Read the AAS Core preview guide → [安装/使用命令见下方] The reusable SKILL.md playbooks, specialized plugins, bundles, workflows, and direct installers remain important. They are the content, curation, distribution, and compatibility layers around AAS Core—not competing primary products. This is an independent community project. It is not affiliated with, sponsored by, endorsed by, or authorized by Google. Google, Antigravity, Gemini, and related product names are referenced only to describe compatibility and install targets. The GitHub repository is canonical; the hosted catalog and browser local Workbench are companion discovery and review surfaces, not a hosted control plane. AAS Core: Agent First Preview The agent composes. You control. AAS keeps the stack reproducible. AAS Core gives the repository one product model: Let the agent choose. The local MCP preserves search skills , get skill , compose stack , inspect stack , and diff stack , and adds read only export selection evidence and inspect selection evidence ; Core does not rank, recommend, exclude, or hide skills. Guide capability coverage. MCP session instructions require the agent to evaluate the full project surface—from architecture, domain behavior, data and integrations through testing, security