smart-ask-data

问数端到端编排（native CLI 版）：从候选 KN 选定知识网络，用 bkn object-type 发现对象类与字段，由编排层 LLM 生成 SQL，再由 ontology dataview query 执行取数；最后输出中文结论与口径说明。当用户需要指标、统计、趋势、SQL 取数或数据查询时使用。

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

70%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

Well-structured orchestration skill with a clear sequenced workflow, a useful checklist, and clean progressive disclosure into real reference files. The main weaknesses are repeated restatement of the same delegation/no-token constraints and command forms that stay at the template/placeholder level rather than being copy-paste ready.

Suggestions

State the delegation chain, the 'Never execute ontology CLI directly' rule, and the 'no token' constraint once each in a single constraints section, then reference it instead of repeating them across the call-chain, dependencies, command-form, and 关系 sections.

Add one concrete end-to-end example with realistic (sample) values filled into the placeholders in SKILL.md or prominently in tool-examples.md, so the command forms are copy-paste ready rather than templates.

Collapse the duplicate KN-source/解析 guidance that appears in both the 主流程 checklist and the 知识网络来源 / 步骤约束 sections into one place.

Dimension	Reasoning	Score
Conciseness	Content is dense and domain-specific (not padded with concepts Claude knows), but the delegation chain (smart-data-analysis→smart-ask-data→ontology-core), the 'Never execute ontology CLI directly' rule, and the 'no token / 不出现 --token' note are each restated multiple times, so it could be tightened — matching the 'mostly efficient but some unnecessary repetition' anchor.	2 / 3
Actionability	Concrete native subcommands and syntax are given (bkn get/object-type, dataview query --sql, --user-id), but they are command-form templates with placeholders (<accountId>, <kn-id>, <dataview-id>) and execution is deliberately delegated to ontology-core/references, so they are not copy-paste ready — the 'some concrete guidance but incomplete' anchor fits rather than the fully-executable anchor.	2 / 3
Workflow Clarity	A clear 5-step sequence with a copyable progress checklist, per-step constraints, and a hard result-display gate (必须同时展示 SQL + 结果) plus empty-result handling; the operation is read-only (SELECT/WITH only), so the missing-feedback-loop cap for destructive/batch ops does not apply.	3 / 3
Progressive Disclosure	SKILL.md is a concise overview with a step→reference table linking four real, one-level-deep reference files (kn-resolve.md, schema-discovery.md, sql-execute.md, tool-examples.md), with content appropriately split and easy navigation — matching the clear-overview-with-well-signaled-references anchor.	3 / 3
	Total	10 / 12 Passed

Description

92%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, concrete description with explicit trigger guidance covering both what the skill does and when to use it. The only weakness is trigger overlap with the sibling smart-data-analysis routing skill, which blurs distinctiveness.

Suggestions

Tighten the 'Use when' clause to phrasings that distinguish 问数 (SQL/metric retrieval) from the broader smart-data-analysis router, e.g. emphasize '用 SQL/ontology 取数出结果' rather than generic 指标/统计/趋势.

Add a one-line disambiguator noting this is the data-retrieval specialist invoked after smart-data-analysis routing, so Claude does not select it for non-retrieval analysis tasks.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — "选定知识网络", "发现对象类与字段", "生成 SQL", "执行取数", "输出中文结论与口径说明" — matching the anchor for several specific concrete actions rather than vague language.	3 / 3
Completeness	Clearly answers both what (the full KN→schema→SQL→execute→summarize pipeline) and when (explicit "当用户需要...时使用" clause), satisfying the both-what-and-when anchor.	3 / 3
Trigger Term Quality	Explicit trigger terms a user would naturally say — "指标、统计、趋势、SQL 取数或数据查询" — give good coverage of natural phrasings, matching the high-coverage anchor.	3 / 3
Distinctiveness Conflict Risk	The 问数 niche and ontology/bkn/dataview toolchain are distinctive, but broad triggers like 指标/统计/趋势/数据查询 overlap with the sibling smart-data-analysis router, so it could still trigger the wrong skill — the 'somewhat specific but could overlap' anchor fits better than the clear-niche anchor.	2 / 3
	Total	11 / 12 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 1 missing, 6 suspicious	Warning

	Total	14 / 16 Passed

Repository: UnicomAI/wanwu
Path: configs/microservice/bff-service/configs/agent-skills/ontology/smart-ask-data/SKILL.md
Commit: f618458

Reviewed: about 13 hours ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.