management-deep-dive

AI Berkshire skill: 管理层纵深研究：买股票就是买人. Source: skills/management-deep-dive.md.

Quality

50%

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

Fix and improve this skill with Tessl

tessl review fix ./codex-skills/management-deep-dive/SKILL.md

Quality

Content

77%

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-sequenced, highly actionable management-research workflow with concrete tables, data sources, a verification command, and a defined output format. Its main weaknesses are verbosity from embedded philosophy/rationale and a monolithic structure with no progressive disclosure into reference files.

Suggestions

Move the embedded philosophy quotes and the "设计理念" rationale into a short reference file (e.g. references/principles.md), keeping only the one-line "关键原则" in SKILL.md to tighten conciseness.

Split the detailed scoring frameworks (兑现率统计, 资本配置评分, 段永平三问) and the report template into a references/ file linked once from the relevant step, reducing the SKILL.md to an overview plus the per-step tables.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient — structured tables, named data sources, and a concrete tool command — but is padded with philosophy quotes (段永平/巴菲特/李录) and explanatory asides ("AI无法和管理层吃饭...") that could be trimmed. Not level 3 because the rationale/quote padding does not earn every token it spends.	2 / 3
Actionability	It provides concrete, specific guidance: fill-in tables for each analysis dimension, exact data sources (股东信/电话会/采访), the command `tools/financial_rigor.py verify-valuation`, numeric scoring thresholds (e.g. 兑现率 >80%), and a defined output path `reports/{公司名}-management-{YYYYMMDD}.md`. For an instruction-only research skill this is fully actionable and copy-paste ready.	3 / 3
Workflow Clarity	A clear 9-step sequence (第一步–第九步) with sub-steps, scoring rubrics acting as checkpoints, an explicit verification command, and a defined report structure. Not capped at 2 because this is a research workflow (not destructive/batch), so absent retry-loops are not penalizing.	3 / 3
Progressive Disclosure	Sections are well-organized, but the skill is a ~290-line monolithic SKILL.md with no bundle/reference files and no one-level-deep references — detailed tables, scoring frameworks, and the report template are all inline. Matches anchor 2 ("content that should be separate is inline"); not level 1 because organization is good, and not level 3 because nothing is split out.	2 / 3
	Total	10 / 12 Passed

Description

22%

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 essentially a skill title plus a source citation rather than a capability statement: it names the domain but lists no concrete actions and provides no usage triggers. This makes it weak on completeness and specificity despite a reasonably distinct niche.

Suggestions

Rewrite the description to lead with concrete actions in third person, e.g. "Conducts in-depth management-team research for an investment target: tracks promises vs. delivery, scores capital-allocation decisions, and evaluates governance."

Add an explicit trigger clause such as "Use when evaluating a company's management team, or when standard investment research leaves the management rating uncertain (★★★ or below)."

Drop the non-trigger metadata ("AI Berkshire skill:", "Source: skills/management-deep-dive.md.") so the description reads as capability + trigger rather than a label.

Dimension	Reasoning	Score
Specificity	The description names the domain ("管理层纵深研究：买股票就是买人") but states zero concrete actions — it is a topic label plus a slogan and a source-file path rather than capability verbs. Not level 2 because anchor 2 requires named actions (e.g. "Processes PDF files and extracts content"); none are present.	1 / 3
Completeness	It gives only a weak "what" (a bare title with no actions) and entirely lacks any "when"/trigger guidance (no "Use when..." clause). This matches anchor 1 ("both very weak"); the cap-at-2 guideline does not rescue it because the "what" itself carries no concrete actions.	1 / 3
Trigger Term Quality	It includes some relevant keywords a user might say ("管理层", "研究", "买股票") but presents them as a title/slogan and adds non-trigger noise ("Source: skills/management-deep-dive.md."), while omitting common variations like "评估管理层" or "CEO评估". Not level 3 because the natural trigger coverage is partial and not phrased as user intent.	2 / 3
Distinctiveness Conflict Risk	It targets a specific niche (management-team evaluation for investing) that is unlikely to trigger unrelated skills, but the title-only phrasing and its role as a "deep-dive version" of a broader investment-research workflow create overlap risk with the parent skill. Not level 3 because the triggers are not distinct enough to fully separate it.	2 / 3
	Total	6 / 12 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: xbtlin/ai-berkshire
Path: codex-skills/management-deep-dive/SKILL.md
Commit: 634a399

Reviewed: 1 day 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.