CtrlK
BlogDocsLog inGet started
Tessl Logo

bkn-creator

【首选技能】凡涉及知识网络的任何操作,优先使用 bkn-creator,不要直接触发 create-bkn。 bkn-creator 是 BKN 全生命周期管理器,负责流程路由识别、阶段门禁、子流程编排与结果回执。 覆盖新增、查找、更新、删除(CRUD)并采用渐进式执行。 触发词:创建知识网络、新建BKN、BKN建模、本体建模、对象类、关系类、动作类、 概念组、BKN文件、BKN push、BKN pull、对象类绑定、关系类映射、 对象类提取、关系类提取、实体关系抽取、 知识网络查询、知识网络更新、知识网络删除。 不应触发:仅查询平台功能、Agent 对话、Vega/Catalog 操作、 健康巡检、数据源连接器等非 BKN 生命周期操作时,应由 ontology-core 等其他技能处理。

66

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./configs/microservice/bff-service/configs/agent-skills/ontology/bkn-creator/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

63%Weight 40%Scale 1-5

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

The body is a well-designed orchestrator overview with genuinely good routing tables, gating rules, and failure handling, and it is token-efficient. Its effectiveness is undermined by the bundle: nearly all executable detail is delegated to COMMON_RULES.md and the five FLOW_*.md files, none of which are present, leaving the skill unable to actually advance past the routing gate.

Suggestions

Ship the referenced files: add COMMON_RULES.md and FLOW_CREATE/FLOW_EXTRACT_TYPES/FLOW_READ/FLOW_UPDATE/FLOW_DELETE.md to the bundle, or inline their essential gating and stage semantics into SKILL.md, so the delegated execution layer is actually reachable.

Remove duplication: the five FLOW_*.md paths appear in both the 流程路由 table and the 子流程调用规则 section, and the COMMON_RULES.md pointer is repeated in three sections — consolidate into a single authoritative reference point.

Add a concrete template for the routing-confirmation output (the exact 回显 format with route, keywords matched, and confirmation prompt) so the top-level gate is fully executable even before the FLOW files are loaded.

DimensionReasoningScore

Conciseness

The body is lean: routing and responsibility tables, terse MUST/Never directives, and no explanations of concepts Claude already knows. It is not a 5 because of avoidable duplication — the five FLOW_*.md targets are listed twice (流程路由 table and 子流程调用规则) and the pointer to ./COMMON_RULES.md is repeated across three sections.

4 / 5

Actionability

The routing layer is concrete (keyword-to-flow table, gating semantics, worked 正例/反例 dialogs, failure-fallback rules), but every executable step delegates to ./COMMON_RULES.md and the ./FLOW_*.md files, none of which exist in the bundle — the actual stage-by-stage instructions are unreachable. This matches the 'concrete guidance but incomplete; missing key details' anchor rather than the minor-gaps anchor at 4.

3 / 5

Workflow Clarity

A clear three-layer sequence (route → confirm gate → discover/preview/confirm/execute/verify/report) with explicit checkpoints: "MUST 回显流程路由识别结果并等待用户确认", "任何写操作必须先完成确认步骤", and failure rollback with no auto-retry. It falls short of 5 because the per-flow stage details and validation specifics are deferred to the missing FLOW_*.md/COMMON_RULES.md files, leaving the checkpoints within each flow implicit.

4 / 5

Progressive Disclosure

The design intent is good — a compact overview signaling one-level-deep references (./FLOW_*.md, ./COMMON_RULES.md, ./references/DOMAIN_ROUTING.md, ./references/common/generic_extraction.md) — but scoring against the actual bundle, six of the core referenced files (COMMON_RULES.md and all five FLOW_*.md) are absent, so the primary navigation paths dead-end. Unreferenced bundle files (bkn_report_template.html, trigger-test-set.jsonl) further blur the structure, fitting the 'some structure but could be better organized' anchor.

3 / 5

Total

14

/

20

Passed

Description

95%Weight 40%Scale 1-5

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 description: it states what the skill does across the full CRUD lifecycle, gives an exhaustive list of natural trigger terms with synonyms, and explicitly fences off sibling skills with anti-triggers. The only weakness is that its capability verbs (gating, orchestration, routing) stay at the orchestration altitude without naming concrete artifacts or outputs.

DimensionReasoningScore

Specificity

The description names the domain (BKN/knowledge-network lifecycle) and lists several concrete capabilities — "流程路由识别、阶段门禁、子流程编排与结果回执" plus "覆盖新增、查找、更新、删除(CRUD)" and "渐进式执行". It stops short of a 5 because the actions are orchestration-level abstractions and no concrete deliverables (e.g., the .bkn file tree) are named, but it is well above the 1-2 concrete actions of a 3.

4 / 5

Completeness

Both questions are answered explicitly: 'what' via "BKN 全生命周期管理器,负责流程路由识别、阶段门禁、子流程编排与结果回执" and 'when' via the concrete 触发词 list plus 不应触发 exclusions. This matches the anchor for clearly and explicitly answering both with concrete trigger phrases; a 4 would require the 'when' to be less explicit.

5 / 5

Trigger Term Quality

The explicit "触发词:" list provides comprehensive natural-language coverage with synonyms and variations — 创建知识网络/新建BKN, 对象类提取/关系类提取/实体关系抽取, BKN push/BKN pull, 知识网络查询/更新/删除. It also adds negative triggers ("不应触发"), matching the comprehensive-synonyms anchor.

5 / 5

Distinctiveness Conflict Risk

The description carves out a clear niche and actively manages conflicts: "优先使用 bkn-creator,不要直接触发 create-bkn" and delegates non-BKN operations ("应由 ontology-core 等其他技能处理") with explicit anti-triggers. Minimal conflict risk, matching the top anchor.

5 / 5

Total

19

/

20

Passed

Validation

100%

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

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository
UnicomAI/wanwu
Reviewed

Table of Contents

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.