harness-engineering

Harness Engineering 编排器——为 AI Agent 构建运行环境规范。通过触发词路由到两个子 skill：harness-generating（规范文档生成/修正）和 harness-gardening（文档一致性巡检与修复）。 Use this skill whenever the user mentions Harness Engineering, AI Agent 运行环境, 驾驭工程, Agent 规范, 上下文工程, 架构约束, 熵管理, 工具能力, 执行与验证, 生成 Harness 文档, Agent 治理, AI 治理规范, 构建 Agent 环境, Agent harness, AGENTS.md, 词汇表, glossary, 文档园艺, 文档巡检, 检查文档一致性, 文档是否过时, harness gardening, 扫描文档, 园艺扫描, 开发地图, dev map, 生成开发地图, 更新开发地图, 项目地图, 代码索引, 源文件索引, 模块索引, 模块依赖, or any workflow involving Harness spec generation or document maintenance.

Quality

82%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Failed to scan

The risk profile of this skill

Quality

Content

78%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.

A well-structured orchestrator body: concise routing tables, concrete mode/param guidance, an explicit ambiguity checkpoint, and excellent progressive disclosure via a verified reference table. Slight headroom on conciseness (the philosophy intro) and on actionability/workflow_clarity because execution is delegated to sub-skills.

Suggestions

Tighten or shorten the '核心理念' paragraph to lift conciseness toward 5 — the routing tables already convey the skill's purpose.

Optionally inline one concrete example of a routed decision outcome (e.g., a sample user phrase → chosen sub-skill + mode + target) to push actionability higher without delegating everything.

Dimension	Reasoning	Score
Conciseness	The body is lean — routing tables and a reference table carry the load with minimal padding — but the '核心理念' paragraph philosophizes ('不是优化模型本身，而是通过系统级工具链解决...') slightly beyond what Claude needs, the minor over-explanation that keeps it off a 5.	4 / 5
Actionability	Concrete, actionable routing guidance: numbered decision rules, mode tables with 判定条件, and explicit params to pass (mode/target/workspace_root); not a 5 because actual execution is delegated to the sub-skills rather than fully specified here.	4 / 5
Workflow Clarity	The routing workflow is clearly sequenced (1 维护类 → 2 生成类 → 3 无法判定 → 询问用户意图) with an explicit ambiguity checkpoint and a 'must read sub-skill SKILL.md' gate; not a 5 because the body delegates the real work, so there is no end-to-end validated process in this file.	4 / 5
Progressive Disclosure	The body is a clear overview that delegates detail to two sub-skills and a well-signaled one-level-deep reference table (file → 用途 → 读取方); all referenced bundle files (references/, scripts/, assets/) were verified to exist, matching the top anchor for easy navigation.	5 / 5
	Total	17 / 20 Passed

Description

86%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 orchestrator description: it clearly states the what (route to harness-generating/gardening) and provides an exhaustive bilingual trigger list with a 'Use this skill whenever...' clause. Its only weakness is that capability specificity is framed as routing rather than direct operations.

Dimension	Reasoning	Score
Specificity	Names the two concrete routing paths with their sub-skills' purposes ('规范文档生成/修正', '文档一致性巡检与修复'), but stays at orchestrator altitude rather than comprehensively listing the operations it performs; not a level-4 'several specific actions' set because the actions are framed as delegation.	3 / 5
Completeness	Explicitly answers 'what' ('为 AI Agent 构建运行环境规范...路由到两个子 skill') and 'when' ('Use this skill whenever the user mentions...') with concrete trigger phrases, hitting the top anchor for both.	5 / 5
Trigger Term Quality	Comprehensive bilingual coverage including synonyms and file artifacts ('Harness Engineering', 'AI Agent 运行环境', 'AGENTS.md', '词汇表/glossary', '开发地图/dev map', 'harness gardening', '文档巡检'), matching the anchor for exhaustive natural-term coverage.	5 / 5
Distinctiveness Conflict Risk	The Harness/AGENTS.md/dev-map niche is mostly distinct with minimal conflict risk, but broad triggers like '上下文工程', '架构约束', and '模块依赖' create minor overlap with adjacent engineering skills, keeping it just below a 5.	4 / 5
	Total	17 / 20 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: TencentBlueKing/bk-bcs
Path: .agents/skills/harness-engineering/SKILL.md
Commit: 99ba07d

Reviewed: 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.