CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

multi-expert-analyzer1

召唤多领域专家智能体,对任何棘手、跨学科、有争议或开放性的问题进行并行深度分析,再经过事实核查员与红队子智能体的克制审阅,最终重写为一篇面向小白、第一人称、图文并茂、有人味的深度长文。触发条件:用户提出任何值得深挖的复杂问题,尤其是跨领域、需要多方视角、需要权威数据支撑的问题;提到"帮我深度分析"、"多专家视角"、"专家怎么看"、"这个问题涉及哪些领域"、"帮我搞清楚 XX 到底是怎么回事"、"用第一性原理分析一下"、"这个观点站得住脚吗"、"帮我论证一下",或者只是抛出一个开放性问题如"为什么 XX 会发生"、"如何看待 XX 现象"、"XX 到底靠不靠谱"。即使用户没有明确要求"专家"或"深度分析"这几个字,只要问题本身跨越多个专业领域、需要审慎论证而非泛泛而谈,也应使用本 skill。本 skill 与仓库中其他类似名称的技能相互独立,请完全按照本文件的流程执行,不要套用其他技能的模板或输出结构。

67

Quality

84%

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

SKILL.md
Quality
Evals
Security

Quality

Content

73%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 instruction skill: the four-phase workflow is unambiguous, every phase has concrete file outputs, and both expert and final-draft self-check loops are explicit. The main cost is repetition — the SVG storage rule and the 'restraint' principle are each stated three times, inflating length without adding information.

Suggestions

State the SVG storage rule once (in 目录约定) and reference it from 阶段二 and 阶段四 with a single phrase like 'SVG 规则见目录约定' instead of repeating the full rule three times.

Consolidate the '克制' principle: state it once at the top of 阶段三 and drop the re-statements in the fact-check, red-team, and reflection subsections, which each re-derive the same idea.

Trim the parenthetical justifications inside the numbered expert-draft requirements (e.g., the multi-sentence rationale after requirement 7) — the requirement statement plus one short reason is enough.

DimensionReasoningScore

Conciseness

The body is mostly efficient instruction-giving with no explanations of concepts Claude already knows, but it repeats itself: the SVG-storage rule appears three times (directory conventions, expert-draft requirement 3, and the final-draft bullet) and the '克制' principle is restated across phases. The repetition goes beyond the 'minor instances' of anchor 4, but there is no padding to the level of anchor 2.

3 / 5

Actionability

Concrete guidance throughout: exact file conventions (markdown/00-question-brief.md, expert-<领域简称>-draft.md, review-fact-check.md, final-<主题slug>.md), a directory tree, an example persona ('有 15 年经验的半导体产业分析师'), 8 numbered per-draft requirements, and sample reflective sentences. A few directives remain judgment-based ('真正理解之后'), keeping it just below fully copy-paste-ready.

4 / 5

Workflow Clarity

The four phases are clearly sequenced with mandatory per-phase disk outputs, an expert self-verification loop (阶段二 requirement 8), and a final self-check checklist with fix-and-recheck ('发现不满足的地方就地微调'). This matches the anchor-5 pattern of explicit validation steps, feedback loops, and a checklist; no destructive/batch cap applies.

5 / 5

Progressive Disclosure

No bundle files exist and the single ~95-line body is well organized with clean section headers, so nothing requires external references. It sits just above the 'simple skill under 50 lines' case — the repeated SVG/克制 guidance is what keeps it from a perfect structure score rather than any missing or buried references.

4 / 5

Total

16

/

20

Passed

Description

91%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: the multi-stage capability is stated concretely, and the trigger section is unusually thorough with quoted natural-language phrases plus a catch-all for implicit cases. The only weaknesses are mild buzzword filler ('有人味的') and a broad 'any open question' trigger surface that could collide with general research skills.

DimensionReasoningScore

Specificity

The description names several concrete pipeline actions ('召唤多领域专家智能体…并行深度分析', '事实核查员与红队子智能体的克制审阅', '重写为一篇面向小白、第一人称、图文并茂…深度长文'), with only minor fuzzy filler like '有人味'. It falls just short of anchor 5 due to slight vagueness, but is well above the 1-2-action anchor.

4 / 5

Completeness

It explicitly answers both 'what' (the full multi-expert → fact-check/red-team → rewrite pipeline) and 'when' via a dedicated '触发条件:' clause with concrete trigger phrases, plus a catch-all rule ('即使用户没有明确要求…也应使用本 skill'). This matches the anchor-5 example structure.

5 / 5

Trigger Term Quality

Comprehensive natural user phrasing is quoted verbatim: '帮我深度分析', '多专家视角', '专家怎么看', '帮我搞清楚 XX 到底是怎么回事', '用第一性原理分析一下', '这个观点站得住脚吗', '帮我论证一下', '为什么 XX 会发生', '如何看待 XX 现象', 'XX 到底靠不靠谱' — covering synonyms and open-question variants a user would naturally say.

5 / 5

Distinctiveness Conflict Risk

It has a clear niche and explicitly disambiguates from similarly-named skills ('本 skill 与仓库中其他类似名称的技能相互独立'), but the trigger surface ('任何…开放性的问题') is broad enough to overlap with general research/deep-analysis skills, keeping it just below anchor 5.

4 / 5

Total

18

/

20

Passed

Validation

87%

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

CriteriaDescriptionResult

relative_links

Relative link issues: 1 missing

Warning

referenced_paths_exist

Referenced path issues: 1 missing

Warning

Total

14

/

16

Passed

Repository

digoal/blog

Path

skills/skills_for_claude_web/multi-expert-analyzer1/SKILL.md

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.