How to Use the Quality-Screen Skill with Seven Hard Indicators for Stock Filtering

The quality-screen skill applies seven non-negotiable financial metrics to filter stocks, instantly excluding companies that fail any hard indicator via the /quality-screen command.

The quality-screen skill is a built-in workflow in the AI Berkshire open-source repository designed for rigorous fundamental analysis. It implements a "7 + 2" rule set that evaluates companies against strict financial thresholds defined in skills/quality-screen.md, ensuring only businesses with durable competitive advantages and clean capital structures proceed to deeper research.

The Seven Hard Financial Indicators

The skill defines seven "hard" indicators that measure capital efficiency, cash generation, debt service, pricing power, and shareholder alignment. According to the source code, any company failing a single metric is immediately excluded unless an exemption rule applies.

Hard Indicator	Exclusion Rule	Measurement Purpose
10-year average ROE	< 8%	Capital efficiency and equity returns
5-year cumulative free cash flow	Negative	Real cash generation versus paper earnings
Interest-coverage ratio (EBIT ÷ Interest)	< 2×	Ability to service debt obligations
Long-term gross margin	< 15%	Pricing power and product differentiation
Operating-cash-flow ÷ Net profit (5-yr avg.)	< 0.7	Earnings quality and cash conversion
Long-term net profit margin	< 5%	Profitability resilience during downturns
5-year total equity dilution	> 20% (excluding M&A)	Management alignment with shareholders

These thresholds are non-negotiable. The skill operates on a "no false-positives" philosophy—it prefers missing a potential winner over retaining a clear loser.

How the Quality-Screen Workflow Executes

The skill processes requests through three distinct stages implemented in the runtime:

1. Input parsing – Determines whether you are filtering a single ticker, an industry, an index, or a thematic list.
2. Parallel data collection – Background agents fetch required financial figures (ROE, FCF, interest coverage, etc.) from company filings and financial data platforms.
3. Rule evaluation – Each company is checked against the seven hard indicators plus two optional exemption rules. The output includes pass/fail status, numeric values, and aggregate statistics like pass-rate and industry ranking.

Invoking the Quality-Screen Skill

The skill exposes a slash-command interface accessible via Claude Code or the AI Berkshire runtime. Use /quality-screen followed by either comma-separated tickers or natural-language descriptions for batch analysis.

Single-stock mode:

/quality-screen 腾讯, 美团, 英伟达

Batch modes (industry, index, or theme):

/quality-screen 恒生指数成分股
/quality-screen 全球AI算力链
/quality-screen 中国啤酒行业

When processing industries or indices, the skill first searches for the top 10-30 constituent companies, then applies the hard-indicator filter to each member in parallel.

Output Format and Examples

The skill returns a markdown table displaying each ticker’s performance against all seven metrics, with ✅/❌ indicators and a final conclusion column.

Example: Single-stock filtering

/quality-screen 腾讯, 美团, 英伟达

| 股票 | ROE (10yr) | FCF 5yr | 利息覆盖 | 毛利率 | 运营/净利 | 净利率 | 股本膨胀 | 结论 |
|------|------------|---------|----------|--------|-----------|--------|----------|------|
| 腾讯 | 12% ✅ | 45B ✅ | 3.8× ✅ | 43% ✅ | 1.2 ✅ | 25% ✅ | 7% ✅ | ✅ Pass |
| 美团 | 6% ❌ | 2B ✅ | 5.1× ✅ | 31% ✅ | 1.1 ✅ | 8% ✅ | 12% ✅ | ❌ Fail (ROE) |
| 英伟达 | 18% ✅ | 12B ✅ | 6.5× ✅ | 68% ✅ | 1.4 ✅ | 30% ✅ | 5% ✅ | ✅ Pass |

Example: Index screening

/quality-screen 恒生指数成分股

> **Summary** – 38% of constituents pass the hard-indicator screen.
> **Top-ranked passes** – 香港电讯、建设银行、汇丰控股.

| 股票 | ROE (10yr) | FCF 5yr | 利息覆盖 | 毛利率 | 运营/净利 | 净利率 | 股本膨胀 | 结论 |
|------|------------|---------|----------|--------|-----------|--------|----------|------|
| 香港电讯 | 9% ✅ | 3B ✅ | 4.2× ✅ | 22% ✅ | 0.9 ✅ | 6% ✅ | 4% ✅ | ✅ Pass |
| 某银行 | 5% ❌ | -0.5B ❌ | 1.8× ❌ | 15% ✅ | 0.6 ❌ | 2% ❌ | 22% ❌ | ❌ Fail (ROE, 利息覆盖, 运营/净利) |

Source Files and Implementation

The quality-screen skill is defined across several key files in the repository:

• skills/quality-screen.md – Canonical skill definition containing the hard-indicator table, exclusion rules, and execution flow.
• codex-skills/quality-screen/SKILL.md – Generated Codex artifact loaded by the runtime at execution time.
• codex-prompts/quality-screen.md – Prompt wrapper that forwards user requests to the skill engine.
• README.md (Quality screen section) – User-facing documentation describing the /quality-screen command syntax and use cases.

These files collectively implement the "7 + 2" rule architecture, where seven hard filters eliminate poor candidates and two exemption rules allow strategic exceptions for special cases like high-growth investment phases.

Summary

• The quality-screen skill uses seven hard financial indicators to filter stocks with zero tolerance for false positives.
• Metrics include 10-year ROE (>8%), 5-year cumulative FCF (positive), interest coverage (>2×), gross margin (>15%), cash conversion (>0.7), net margin (>5%), and equity dilution (<20%).
• Invoke via /quality-screen with comma-separated tickers or natural-language industry/index descriptions.
• The workflow parses input, collects data in parallel, and evaluates against non-negotiable thresholds defined in skills/quality-screen.md.
• Output includes detailed pass/fail tables with numeric values and summary statistics for batch screenings.

Frequently Asked Questions

What are the seven hard indicators in the quality-screen skill?

The seven hard indicators are: 10-year average ROE (minimum 8%), 5-year cumulative free cash flow (must be positive), interest-coverage ratio (minimum 2×), long-term gross margin (minimum 15%), operating-cash-flow to net profit ratio (minimum 0.7), long-term net profit margin (minimum 5%), and 5-year equity dilution (maximum 20%). Each metric targets a specific dimension of business quality, from capital efficiency to shareholder alignment.

How do the exemption rules work in the quality-screen skill?

The skill implements a "7 + 2" architecture where two exemption rules can override hard-indicator failures. For example, a young, high-margin company in a strategic investment phase might receive an exemption despite negative short-term cash flows. These exceptions prevent the elimination of genuine high-quality businesses undergoing temporary transitions while maintaining strict standards for mature companies.

Can I screen entire indices or industries with the quality-screen skill?

Yes. The skill accepts natural-language inputs like "恒生指数成分股" (Hang Seng Index constituents) or "中国啤酒行业" (China beer industry). It automatically searches for the top 10-30 relevant companies, then runs the seven hard-indicator filter on each constituent in parallel, returning aggregate statistics like pass-rate and industry rankings alongside individual ticker results.

Where is the quality-screen skill logic defined in the AI Berkshire repository?

The canonical definition resides in skills/quality-screen.md, which specifies the hard-indicator thresholds and exclusion rules. The runtime implementation loads from codex-skills/quality-screen/SKILL.md, while codex-prompts/quality-screen.md handles command routing. User documentation appears in the main README.md file under the Quality screen section.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}