CtrlK
BlogDocsLog inGet started
Tessl Logo

harness-gardening

文档园艺——持续维护 harness-engineering 产出的文档与项目实际状态的一致性。 支持 PR 轻量检查(代码变更后按需触发受影响维度)、全量深度扫描(十二维度完整检测) 和定向检测(targeted,指定单个组件/维度进行针对性巡检)。 完整扫描后:若存在 P1,向用户列出方案并请求确认,确认后与 P0 一并修复并提交;仅有 P0 时自动修复提交。 由 harness-engineering 按维护类触发词路由调用,不独立触发。

62

Quality

73%

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 ./.agents/skills/harness-engineering/harness-gardening/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

71%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 content is well-structured, actionable, and exemplary in progressive disclosure, with clear workflow sequencing and real verification checkpoints. Its main weakness is repetition: the disposal/commit/last-commit sub-procedure is duplicated across three mode sections instead of being factored out once.

Suggestions

Extract the shared P0/P1 disposal + commit + last-commit write-back procedure into a single referenced block that all three modes invoke, to remove the triplicated sub-steps and lift conciseness.

Add an explicit error-recovery branch for when a dimension detection or git command fails mid-flow, so the workflow has a feedback loop beyond the final checklist.

DimensionReasoningScore

Conciseness

The body is dense and operational with no padding of concepts Claude already knows, but the P0/P1 disposal, commit, and last-commit write-back sub-steps are repeated nearly verbatim across the PR, full, and targeted sections, which could be tightened into a shared block.

3 / 5

Actionability

It provides concrete, executable commands — `eval $(bash "$SKILL_ROOT/.../harness-setup-git.sh")`, `git rev-parse HEAD`, `git add`, `git commit --amend --no-edit` — with specific env-var outputs and paths, with only minor gaps where dimension rules are deferred to references.

4 / 5

Workflow Clarity

Each mode is a clearly numbered sequence with checkpoints (P1 human confirmation, per-step TodoWrite completion, and the 清单验收 verification loop that re-executes pending items), so the destructive/batch cap at 3 does not apply; minor gaps remain around error recovery when a dimension detection itself fails.

4 / 5

Progressive Disclosure

The body is an overview pointing to one-level-deep, well-signaled references — a 参考资源 table with a '何时读取' column maps each file (verified: references/detection-dimensions.md and references/output-templates.md exist) to its use point, with no nested reference chains.

5 / 5

Total

16

/

20

Passed

Description

76%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.

The description is specific, distinctive, and answers both what it does and when it runs, but its trigger-term quality is generic ('维护类触发词') without the concrete phrases users would naturally say, capping completeness below the top level. It is written in proper third-person voice with no first/second-person phrasing.

Suggestions

Enumerate concrete maintenance trigger words a user might say (e.g., '文档过时', 'gardening', '巡检文档一致性') instead of the generic phrase '维护类触发词'.

Tighten the 'when' clause into an explicit 'Use when...' style trigger listing to push completeness from 4 to 5.

DimensionReasoningScore

Specificity

Names the domain (文档园艺) and lists multiple concrete actions — PR lightweight checks, full twelve-dimension deep scan, targeted detection, P0 auto-fix-and-commit, and P1 list-confirm-fix-commit — giving comprehensive coverage.

5 / 5

Completeness

Both 'what' (three detection modes plus P0/P1 disposal) and 'when' ('由 harness-engineering 按维护类触发词路由调用,不独立触发') are stated, but the 'when' is expressed as parent-skill routing rather than concrete user trigger phrases, so it could be more explicit.

4 / 5

Trigger Term Quality

It references '维护类触发词' (maintenance trigger words) but never enumerates the natural phrases a user would actually say, so keyword coverage has relevant terms yet misses common variations and synonyms.

3 / 5

Distinctiveness Conflict Risk

It carves a clear niche — doc-consistency gardening for harness-engineering output, routed only by the parent skill — with distinct triggers and minimal overlap risk with other skills.

5 / 5

Total

17

/

20

Passed

Validation

93%

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

Validation15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository
TencentBlueKing/bk-bcs
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.