SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsearnings-team

earnings-team

Writing & research

View repository

Updated 10-16 01:56

About this skill

Codex adapter note

This skill is generated from skills/earnings-team.md so Claude Code and Codex users share one canonical workflow.

  • Treat $ARGUMENTS as the user's request in the current Codex thread.
  • When the source mentions Claude-only surfaces such as Task, Agent, WebSearch, Bash, Read, or Write, use the closest Codex capability available in this session: subagents when available, web search when needed, shell commands for local tools, and normal file edits for workspace files.
  • Use shared project tools from tools/ in this repository. Prefer running commands from the repository root with paths like python3 tools/financial_rigor.py ...; if the current thread starts outside the repo, locate the actual checkout path first instead of assuming a fixed home-directory path.
  • Before starting research, run the date command to confirm today's date; treat it as the baseline for "latest" data and state the data cutoff date in the report header. Never assume the current date from training data.
  • Preserve the research quality rules from AGENTS.md: cross-check financial data, use exact arithmetic tools for valuation/math, and clearly label uncertainty and source gaps.
Financial Report Deep-Read Team: Four Masters Parallel Interpretation + Official Account Publishing

Conduct a team-based financial report deep-read analysis of $ARGUMENTS. Four masters interpret the financial report in parallel, an editor polishes it into an article, reader reviewers ensure quality, and the final output is a ready-to-publish official account article.

Supported input formats：公司名 季度, for example:腾讯 2025Q4、PDD 2025年报、美团 最新

Design Philosophy

A good financial report analysis must solve two problems:

  1. You can understand the future yourself—this requires in-depth research from four different perspectives
  2. Readers can understand the value—this requires editorial polishing and quality control from the reader's perspective

The workflow of this Skill is divided into three stages:

  • Stage 1 · Research: Four masters conduct parallel close reading of the financial report (Duan Yongping examines the essence of the business, Buffett scrutinizes financial quality, Munger reads changes in competition, Li Lu hunts for risk signals)
  • Stage 2 · Synthesis: The Team Lead integrates the four perspectives and produces the first draft of the research report
  • Stage 3 · Publishing: Editor Agent rewrites it into an official account article + Reader Review Agent proposes revisions → Team Lead finalizes

Stage 1: Four Masters Parallel Research

Step 1: Obtain first-hand materials

Use the Agent tool to launch background Agents in parallelto obtain the following source materials:

Material Type Source Priority
Original financial report Company IR page, SEC EDGAR (US stocks), HKEXnews (Hong Kong stocks), CNINFO (A-shares) Highest
Earnings call transcript Seeking Alpha, company IR page, Xueqiu Highest
Management letter to shareholders Extracted from the annual report High (annual reports only)
Previous financial report/earnings call Same as above High (for commitment tracking)

Material Availability Rating：

Level Characteristics Impact
Grade A Complete original text obtained Execute all steps normally
Grade B Only partial original text or third-party summaries obtained Mark "non-original source" and reduce the weight of notes analysis
Grade C Only news reports and data website summaries available Focus on core data changes, skip notes mining, mark "insufficient first-hand materials"

Inform each Agent of the material availability rating, affecting their depth of analysis.

Step 2: Show the team framework to the user

Stage Role Master/Positioning Core Task
Research Team Lead(yourself) Overall Coordinator Coordinate, synthesize, finalize
Research Business Essence Interpreter Duan Yongping Has this business gotten better or worse?
Research Financial Quality Auditor Buffett Is it real money or fake money that's being made?
Research Interpreter of competitive change Munger How is the competitive landscape changing?
Research Hunter of risk signals Li Lu What is management hiding?
Publish Edit Official account writing Rewrite research reports into good articles
Publish Reader review Ordinary investors Can readers understand it? Is it rewarding?

Step 3: Launch 4 parallel research Agents

Use the Agent tool inthe same messageto launch 4 background Agents.

Agent 1: Interpretation of the essence of the business (Duan Yongping's perspective)

Core question: Has the essence of the business reflected in this financial report improved or worsened?

Duan Yongping: "Investing is buying a business. Looking at a financial report is not about looking at numbers; it is about seeing whether this business has changed."

Analysis content:

  1. Decomposition and interpretation of revenue structure

    • Revenue by business/region, which are accelerating and which are decelerating
    • Not just listing numbers—what business logic does each business segment reflect
    • Does revenue growth come from "volume" or "price"? Which is healthier?

  2. Changes in user/customer value

    • Changes in operating metrics such as DAU/MAU/paid users
    • Quality metrics such as user time spent, ARPU, and retention rate
    • Is the platform/product's value to users strengthening or weakening?

  3. Moat check

    • Changes in gross margin reflect whether pricing power is solid
    • Changes in market share reflect whether competitive barriers are effective
    • Signals that customer switching costs/network effects have been weakened

  4. Evaluation of "good business" standards

    • Duan Yongping's three conditions: differentiation, pricing power, sustainable competitive advantage—changes in this period
    • Is the business becoming "heavier" or "lighter"?
    • If the company closed tomorrow, would users be in great pain? Has that changed because of this financial report?

  5. Management's product intuition

    • When discussing products/users, does management use concrete language or bureaucratic language
    • Whether there are impressive product insights or worrying signals of disconnection

Output requirements: Mark each subitem as 🟢improving / 🟡flat / 🔴deteriorating, and give a Duan Yongping-style summary comment.

Agent 2: Financial quality audit (Buffett's perspective)

Core question: Is this company making real money or fake money? Has the margin of safety changed?

Buffett: "When I look at every financial report, the first thing I do is turn to the cash flow statement."

Analysis content:

  1. Extraction and verification of core financial data

    • Revenue, gross profit, operating profit, net profit—both GAAP and Non-GAAP
    • GAAP vs Non-GAAP difference: how much, where, and whether the gap is widening or narrowing
    • Cross-verify key data with at least two sources
    python3 tools/financial_rigor.py cross-validate \
  --metric "revenue" --values {值1} {值2} --sources "来源1" "来源2"

  2. Cash flow analysis (most important)

    • Operating cash flow vs net profit ratio (>100% good, <80% caution)
    • Free cash flow = operating cash flow - capital expenditures
    • Composition of capital expenditures: maintenance vs. expansion
    • Amount of buybacks and dividends

  3. Profit quality check

    • Accounts receivable growth rate vs. revenue growth rate
    • Inventory growth rate vs. revenue growth rate
    • Trend of the gap between operating cash flow and net profit
    • Whether capitalized expenditures suddenly increased
    • Proportion of non-recurring gains

  4. Balance sheet health

    • Changes in net cash/net debt
    • Changes in accounts receivable/inventory turnover days
    • Impairment risk of goodwill and intangible assets

  5. Valuation and margin of safety update

    python3 tools/financial_rigor.py verify-market-cap \
  --price {价格} --shares {股本} --reported {报告市值} --currency {币种}
python3 tools/financial_rigor.py verify-valuation \
  --price {价格} --eps {EPS} --bvps {每股净资产}
python3 tools/financial_rigor.py three-scenario \
  --price {价格} --eps {EPS} --shares {股本亿} \
  --growth {乐观} {中性} {悲观} --pe {乐观PE} {中性PE} {悲观PE}

Output requirements: All calculations are accompanied by tool output records, profit quality signal lights 🟢/🟡/🔴, and a Buffett-style summary commentary.

Agent 3: Competitive landscape interpretation (Munger's perspective)

Core question: What changes in the competitive landscape does this financial report reveal?

Munger: "I want to know where I'm going to die, so I don't go there."

Analysis content:

  1. Infer competitive changes from financial report data

    • Revenue growth rate vs. industry growth rate—outperforming or underperforming?
    • Changes in gross margin reflect intensifying/easing competition
    • Changes in marketing expense ratio—do you need to spend more to acquire customers?
    • R&D investment—proactive investment or forced catch-up?

  2. Comparison with competitors in the same period

    • Comparison of key indicators of major competitors in the same period (if already published)
    • Comparison of growth rate, profit margin, and investment intensity
    • Who is winning? Who is losing?

  3. Management's discussion of competition

    • How the competitive environment is described on the earnings call
    • Are competitors named? Is the tone confident or anxious?
    • Are there any new competitive threats?

  4. Industry trend signals

    • Impact of technological change (AI/new platforms, etc.)
    • Impact of regulatory changes on the competitive landscape
    • Consumption/demand-side trends

  5. Munger-style inversion thinking

    • What would kill this company? Does this period's financial report contain signals pointing to these threats?
    • Looking back 5 years from now, will this financial report be a "turning point"?

Output requirements: Competitive landscape judgment (strengthening/flat/deteriorating), competitor comparison table, Munger-style inversion commentary.

Agent 4: Risk signal hunter (Li Lu's perspective)

Core question: What has management concealed in this financial report? Which signals are flashing?

Li Lu: "The most important thing in investing is avoiding permanent capital loss."

Analysis content:

  1. Management tone analysis

    • Read management discussion and earnings call remarks paragraph by paragraph, marking signals:
    • 🟢Candid signal (proactively acknowledging problems) / 🟢Clear signal (has quantified targets)
    • 🔴Vague signal (empty talk) / 🔴Deflection signal (answering off-topic) / 🔴Externalizing attribution

  2. Commitment tracking

    • Specific commitments made by management last period vs. actual fulfillment this period, compared item by item
    • Duan Yongping: "To see whether management is reliable, just look at whether they have done what they said before."

  3. Notes and hidden information

    • Related-party transactions, equity incentive dilution, contingent liabilities
    • Changes in accounting policies, segment margin differences
    • Changes in customer/supplier concentration

  4. Selected Q&A from the earnings call

    • The 3-5 sharpest analyst questions and quality scores for management's answers

  5. Risk of permanent capital loss

    • Whether signals have appeared that could lead to permanent loss
    • New developments in regulatory/compliance/litigation risks
    • Whether management has made irreversible wrong decisions

Output requirements: Management credibility score ★1-5, promise fulfillment rate, risk signal checklist, Li Lu-style summary commentary.

Step 4: Track progress

Show the user in real time:

📊 {公司名} {期间} 财报精读进度
━━━━━━━━━━━━━━━━━━━━━━━
阶段一·研究
  ☐ 段永平·生意本质    ⏳ 分析中...
  ☐ 巴菲特·财务质量    ⏳ 分析中...
  ☐ 芒格·竞争格局      ⏳ 分析中...
  ☐ 李录·风险信号      ⏳ 分析中...
阶段二·合成            ⏸ 等待中
阶段三·发布            ⏸ 等待中

Each time a report is received, update progress and display the core findings (3-5 items).

Phase 2: Team Lead synthesizes the research report

After all 4 research reports are in, the Team Lead synthesizes and produces the first draft of the research report.

Synthesis points—not piecing reports together, but finding intersections and contradictions:

  1. Points of consensus across the four perspectives: Conclusions all four masters agree on, with the highest credibility
  2. Points of contradiction across the four perspectives: For example, Duan Yongping says the business has improved, but Munger says competition is worsening—this kind of contradiction is the most valuable analysis
  3. Neglected corners: Things none of the four focused on—could they actually be the most important?

Research report structure

# {公司名} {期间} 财报精读报告
**四大师并行解读 | {日期}**

## 一、一句话结论
> 50-100字：超/符/低预期，核心变化，对投资论文的影响。

## 二、本期最重要的3个变化
聚焦真正重要的变化，不罗列数据，每个变化100字以内。

## 三、四大师评分表
| 视角 | 大师 | 核心问题 | 结论 | 评分 | vs上期 |
|------|------|---------|------|------|--------|

## 四、核心数据速览
关键财务和运营指标表格（本期 vs 上期 vs 同比）

## 五、各视角深度分析
每个视角3-5条最重要发现

## 六、管理层语气与承诺追踪
承诺兑现表 + 语气变化分析

## 七、四大师会怎么做？
| 大师 | 如果持有 | 如果没持有 | 理由 |

## 八、结论
1. 超/符/低预期？
2. 投资论文影响：强化/无影响/削弱/破裂
3. 下一个催化剂
4. 操作建议

Phase 3: Editorial polishing + reader review

After the research report is completed,in parallellaunch two Agents:

Agent 5: Editor (rewriting for a WeChat public account article)

Positioning: Rewrite the hardcore research report into an article that WeChat public account readers enjoy and can understand.

Core principles：

  • Preserve all key data and conclusions without reducing professional depth
  • Improve the way things are expressed so that non-professional investors can also follow the logic
  • It is not "popularization," but "making professional content effortless to read"

Specific tasks：

  1. Title and opening

    • The title should be informative and click-worthy, but not clickbait
    • Example of a good title: "Kuaishou spent 26 billion betting on AI—did it win?"
    • Example of a bad title: "Shocking! Kuaishou's earnings blow up!"
    • Within the first 100 characters, make clear: what the most important conclusion of this earnings report is, and why readers should care

  2. Structural optimization

    • A research report is for yourself; a WeChat public account article is for others—adjust the logical order
    • Put the "3 most important changes" at the very front (inverted pyramid structure)
    • Keep tables but streamline them; change long analysis passages into bullet points
    • Insert a "stage summary" about every 500 characters to help readers digest

  3. Expression polishing

    • Explain stiff financial terms with analogies/scenarios: "operating cash flow is 30% lower than net profit" → "earned 100 yuan but only felt 70 yuan in your pocket"
    • The four masters' commentary quotes are the soul of the article—make sure each one reads sharply and is memorable
    • Paragraphs no longer than 4 lines, sentences no longer than 30 characters
    • Use contrast and反差 moderately to create reading rhythm

  4. Reader Value Check

    • For each chapter, ask yourself: after reading this passage, what decision can the reader make? If the answer is "nothing can be done," either rewrite it or delete it
    • The end of the article needs a clear "So what?"—provide operational guidance separately for holders and observers

  5. Format Adaptation

    • WeChat Official Account layout-friendly: short paragraphs, clear subheadings, concise tables
    • Add appropriate dividers and quote formatting
    • Keep the article length between 1,000 and 3,000 Chinese characters (too long and readers will drop off)

Output: the complete rewritten WeChat Official Account article.: the complete rewritten WeChat Official Account article.

Agent 6: Reader Review (Ordinary Investor Perspective)

Positioning: review the article as an ordinary investor who "focuses on value investing, has basic financial knowledge, and holds/follows the company.": review the article as an ordinary investor who "focuses on value investing, has basic financial knowledge, and holds/follows the company."

Review Dimensions：

  1. Readability (weight 30%)

    • How many minutes does it take to read the whole article? Are there paragraphs you want to skip?
    • Which parts are hard to understand or require repeated reading?
    • How is the pacing? Is there a feeling of "getting tired of reading"?

  2. Information Value (weight 30%)

    • After reading, has my understanding of this company deepened?
    • Were there any "oh, so that's how it is" moments?
    • Compared with analyses I've seen elsewhere, what is unique about this piece?
    • Which information is redundant and can be deleted without affecting understanding?

  3. Credibility (weight 20%)

    • Do the data have sources? Are key judgments supported by evidence?
    • Are both sides presented? Or is it only bullish/bearish?
    • Are there judgments that feel "this is too confident" and make people uncomfortable?
    • Are the quotations from the four masters appropriate and powerful?

  4. Action Guidance (weight 20%)

    • After reading, do I know what to do?
    • Are the suggestions for "holders" and "observers" specific enough?
    • What should be watched next? (catalysts, time nodes)

Output Format：

## 读者评审报告

### 总体评分：X/10

### 优点（2-3条）
读者视角下文章做得好的地方

### 必须修改（硬伤）
- 问题1：具体描述 → 建议修改方式
- 问题2：...

### 建议优化（锦上添花）
- 建议1：...
- 建议2：...

### 读者最想知道但文章没回答的问题
- 问题1
- 问题2

### 一句话总评

Team Lead Finalizes

After receiving the editor's rewritten draft and the reader review report:

  1. Handle the "must fix" items from the reader review— revise them one by one
  2. Selectively adopt the "suggested optimization" items— judge whether they are worth it
  3. Add "questions readers want to know but that were not answered"— add them if there is data support
  4. Final read-through— ensure the revised full text is coherent and logically consistent

Output File

reports/{公司名}/
├── {公司名}-earnings-{期间}.md           ← 最终公众号文章（定稿）
├── {公司名}-earnings-{期间}-研究底稿.md   ← 四大师合成研究报告（自用）
├── {公司名}-earnings-{期间}-段永平.md     ← 生意本质解读
├── {公司名}-earnings-{期间}-巴菲特.md     ← 财务质量审计
├── {公司名}-earnings-{期间}-芒格.md       ← 竞争格局解读
├── {公司名}-earnings-{期间}-李录.md       ← 风险信号分析
└── {公司名}-earnings-{期间}-读者评审.md   ← 读者评审报告

Data Spot Check (release process)

Perform a spot check on the final article:

python3 tools/report_audit.py extract \
  --report reports/{公司名}/{公司名}-earnings-{期间}.md

python3 tools/report_audit.py verdict \
  --results '<填好的JSON>' \
  --report {报告文件名}

[Pass] all pass → can be published;[Send Back] any fail → revise and re-review.

Relationship with Existing Skills

Skill Positioning When to Use
/earnings-review Single-Agent Financial Report Close Reading Quick run-through, only one perspective needed
/earnings-team(this Skill) Six-Agent Team Close Reading + WeChat Official Account Publishing Key financial reports of important companies, requiring depth + publishing
/investment-team Four-Agent Comprehensive Company Research Researching a Company for the First Time

Key Principles

  • Read the original text, not summaries: Do everything possible to obtain primary sources
  • The four perspectives are not four departments: They must corroborate and challenge each other, not each speak separately
  • The value of the Team Lead lies in comprehensive judgment: Find intersections and contradictions, not stitch together reports
  • Conclusions must be clear: "Overall, it basically meets expectations but there are also some points worth noting" is not allowed
  • Adversarial testing runs throughout the entire process: Every positive finding comes with counterarguments
  • Editing is not about lowering the level of expertise: It is about making expert content more readable, not turning it into popular science
  • Reader review is not a formality: Really find faults from the reader's perspective
  • Data accuracy: Cross-validate key data, use the financial_rigor.py tool for verification and calculation