使用方法
- 复制安装命令,在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物,并对照来源检查输出。
npx skills add https://github.com/holistics/anfra-skills --skill anfra-aql开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况,不等于独立质量或安全测试。
按源仓库内容呈现,保留标题、案例、代码、表格、链接以及原文引用的演示图片。
AQL core knowledge
AQL is Analytic Querying Language for data analytics. This skill is the knowledge base you consult while authoring AQL. It doesn't run queries — pair it with actionable skills such as [](../anfra-write-aql/), [](../anfra-validate-aql/), etc.
How to use these references
AQL is not SQL — don't write it from general or SQL knowledge. Read the core lessons before writing your first AQL; then pull in the specific per-function docs and examples your task needs (that part is on-demand — you needn't read everything).
- `references/aqlearn.md` — read this first, before writing any AQL. The single, complete AQL lessons + rules + glossary reference (aggregation, dimensions/measures, level of detail, nested aggregation, window functions, ranking, relationships, SQL passthrough, …). Don't skip it — it holds the vital details you won't find by guessing, and reading it up front is faster than writing wrong AQL and fixing it.
- `references/aql/` — per-function reference docs. Look up the exact
syntax, semantics, and argument order for a specific function or operator. Files are named by function/operator (e.g. sum.md, rank.md, where.md); browse the folder and open the one you need. Use this to double-check — never assume a function, argument, or argument order.
- `references/examples/` — worked examples. Follow proven patterns and
avoid known pitfalls.
- Index —
references/examples/INDEX.mdis a table with columns
id | kind | analytic | problems | techniques | gotchas (kind is good or bad; the last three are tag slugs). To find examples, grep the index (or the example frontmatter) for a tag slug, or scan the analytic column; then open the matching references/examples/<id>.md.
- Example file —
references/examples/<id>.mdhas YAML frontmatter
(id, kind, analytic, problems, techniques, gotchas) followed by ## Dataset fields (``yaml), ## AQL (``aql), and learning points.
- Tag glossaries (same folder) define the slugs:
PROBLEMS.md(how to
frame the problem), TECHNIQUES.md (AQL idioms), GOTCHAS.md (pitfalls; entries marked [silent] fail with no validation error — a wrong-but-valid result — so look these up proactively).
- The AQL docs, for what these references don't cover: look it up with
[](../anfra-docs/).
anfra skills 0.21.0