earnings-team

Analyzes earnings reports through four parallel investor perspectives and produces publishable articles.

16.4k|2.5k|Updated Apr 7, 2026
One-click install
npx skills add https://github.com/xbtlin/ai-berkshire --skill earnings-team-xbtlin
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: earnings-team
Source: https://github.com/xbtlin/ai-berkshire/tree/main/codex-skills/earnings-team
Command: npx skills add https://github.com/xbtlin/ai-berkshire --skill earnings-team-xbtlin

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Reading an earnings report from a single angle produces shallow, one-sided analysis that cannot support real investment decisions. This Skill coordinates a six-agent team that reads earnings reports in depth from four value-investing perspectives, synthesizes the findings, and turns them into a polished, publish-ready article. ## Core Features & Use Cases - Four Parallel Research Agents: Duan Yongping analyzes business quality, Buffett audits financial quality and cash flow, Munger reads competitive dynamics, and Li Lu hunts for hidden risk signals and management red flags. - Synthesis and Contradiction Detection: A Team Lead merges the four reports, highlighting consensus points, contradictions between perspectives, and overlooked areas rather than simply concatenating findings. - Editorial and Reader Review Pipeline: An editor agent rewrites the research report into a WeChat-ready article, a reader-reviewer agent scores readability, credibility, and actionability, and the Team Lead finalizes the draft. - Data Rigor and Audit: Financial figures are cross-validated and valuation math is verified with repository tools (financial_rigor.py, report_audit.py), with a pass/fail audit gate before publication. - Use Case: Run it on "Tencent 2025Q4" to receive four perspective-specific research notes, a synthesized research draft, a reader review report, and a finalized article saved under reports/. ## Quick Start Ask the assistant to run the earnings-team analysis on a company and period, for example: analyze PDD's latest annual report with the four-master earnings team and produce a publishable article.

Frequently Asked Questions about earnings-team

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I analyze an earnings report with multiple AI agents?

Provide a company name and period such as "Tencent 2025Q4". The skill launches four parallel research agents covering business quality, financial quality, competitive dynamics, and risk signals, then synthesizes their findings into one report.

What input formats does the earnings-team skill accept?

It accepts a company name plus a reporting period, such as "PDD 2025 annual report" or "Meituan latest". The agents then fetch primary sources like filings, earnings call transcripts, and shareholder letters from IR pages, SEC EDGAR, HKEX, or CNINFO.

How does the skill verify financial data accuracy?

Key figures are cross-validated across at least two sources, and valuation math is computed with the repository's financial_rigor.py tool rather than mental arithmetic. A final report_audit.py check acts as a pass/fail gate before the article is published.

What happens when original filings are not available?

The skill assigns a source availability grade of A, B, or C. With B-grade sources it labels data as non-original and reduces footnote analysis weight; with C-grade it focuses only on core figures and explicitly marks the report as lacking first-hand material.

When should I use earnings-team instead of a single-agent earnings review?

Use earnings-team for key earnings of important companies where you need both deep multi-perspective research and a publish-ready article. Use the lighter single-agent earnings review when you only need a quick read from one perspective.

What output files does the earnings analysis produce?

It writes a finalized article, a synthesized research draft, four perspective-specific notes, and a reader review report into a per-company folder under reports/, named by company and reporting period.