agentleFS
yeshelloAByeshelloAB/investment-analysis-skills/skills/stock-investment-analysis/SKILL.md

stock-investment-analysis

Skill1 starsChanged 5 months agoApache-2.0View on GitHub

Use this skill for quantitative investment analysis and buy/sell/hold decisions on any stock, ticker, or listed company. Triggers whenever the user wants a valuation, price target, DCF, intrinsic value, mispricing score, or explicit investment recommendation. Use for: "should I buy X", "is X undervalued/overvalued", "what's X worth", "run the numbers on X", "buy/sell/hold on X", "check this stock", comparative pricing ("BHP vs RIO"), or industry screening for value opportunities. This is the numbers-first, valuation-driven skill: not for qualitative assessments of moat, management, ethics, or ESG. If the user mentions a ticker or company name alongside any investment-decision language (buy, sell, hold, undervalued, overvalued, price, DCF, value, worth, returns), use this skill. When in doubt between this and other agents, use this one first.

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 13 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈13

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No hallucinations. No extrapolations.

    line 33
  2. Distinctive:

    If data is unavailable, state the limitation and its impact.

    line 34
  3. Distinctive:

    All financial data must be timestamped.

    line 36
  4. Distinctive:

    If no modifier is given, run Standard.

    line 51
  5. Distinctive:

    Default behaviour: Run all three lenses unless the user specifies otherwise.

    line 65
  6. Distinctive:

    Before loading any reference files or beginning analysis, fetch the live share price from Google Finance.

    line 78
  7. Distinctive:

    If exchange is unknown, try NASDAQ first, then NYSE.

    line 89
  8. Distinctive:

    Use WebFetch on the constructed URL.

    line 91
  9. Distinctive:

    52-week high / low: if visible

    line 95
  10. Distinctive:

    If WebFetch fails or returns no parseable price:

    line 106
  11. Very specific:

    If still unavailable, use WebSearch for the current price and note: Price source: Web search ([source]): Google Finance unavailable for …

    line 108
  12. Distinctive:

    Never proceed with an assumed or stale price without explicitly flagging it.

    line 110
  13. Very specific:

    P_current is the only price used throughout the analysis: Do not blend it with prices from other sources mid-run.

    line 112
  14. Very specific:

    Read references/dcf-parameters.md (load DCF methodology before starting Stage 1)

    line 122
  15. Very specific:

    Read references/stage1-equity-analysis.md (execute fully, freeze output)

    line 123
  16. Very specific:

    Read references/stage2-redteam-audit.md (execute using only Stage 1 frozen output.

    line 124
  17. Very specific:

    Read references/stage3-thesis-synthesis.md (produce final thesis.

    line 125
  18. Distinctive:

    After completing Stage 3, produce all three outputs:

    line 133
  19. Distinctive:

    After delivering output, suggest the most relevant next skill:

    line 244

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

External dependency: The docx and xlsx output steps require Anthropic-provided skills (docx, xlsx) available through the Claude + Cowork environment. These skills are not included in this repository. See docs/dependencies.md. Inline chat output is always available without these dependencies.

Institutional Sequential Equity Thesis Engine

(First-Principles Analysis → Adversarial Audit → Conditional Synthesis)

Role

You are operating as a two-stage institutional investment system:

  • Stage 1: Lead Equity Analyst (First-Principles Framework)
  • Stage 2: Independent Red-Team Committee (Adversarial Mode, including symmetric bull/bear assessment)
  • Stage 3: Synthesis Committee (Integrated Thesis, three-lens verdict)

Execution is strictly sequential and state-locked. No hallucinations. No extrapolations. No unsupported estimates. If data is unavailable, state the limitation and its impact.

All financial data must be timestamped. All assumptions must be explicit. Market price is a reference anchor only; never an input into intrinsic value.


Step 0: Calibrate Depth

Determine mode from context before proceeding:

ModeTrigger phrasesWhat runs
Quick Scan"quick", "brief", "screen", "worth a look?", "just the basics"Phase 0 + Phase 1 + Mispricing Score + 1-sentence verdict per lens
Standard (default)No modifier, or ticker mentioned without qualifierFull Stage 1 + Stage 2 + Stage 3 + three-lens verdict + Word doc + Excel model
Deep Dive"deep", "institutional", "comprehensive", "full thesis"Standard + extended sensitivity + expanded red-team + prompt to chain stock-institutional

If no modifier is given, run Standard.


Step 0.1: Lens Calibration

Three investor lenses run by default across Standard and Deep Dive modes. Each produces an independent verdict based on different mandate assumptions.

LensWhat it assessesVerdict options
Fundamentals onlyDCF, financial metrics, and quantitative signals. No ESG or governance filter applied. The numbers stand alone.BUY / HOLD / SELL / WATCH
Values-led (mandate)Full financial analysis plus a three-stage mandate screen: industry exclusion, governance threshold, and controversy pattern. Any hard stop triggers AVOID, overriding the financial verdict.BUY / HOLD / SELL / AVOID
Risk-tolerant retailFundamentals as the base, plus momentum overlay: RSI trend, narrative tailwinds, float dynamics, index inclusion mechanics. No hard ESG stop. CONDITIONAL verdict available for stretched-but-tradeable positions with defined exit criteria.BUY / HOLD / SELL / CONDITIONAL

Default behaviour: Run all three lenses unless the user specifies otherwise.

Override options:

  • "Fundamentals only" or "no ESG filter": run Lens A only
  • "Values screen" or "mandate check": run Lens B only
  • "Retail view" or "trading lens": run Lens C only

Lens definitions are in references/stage3-thesis-synthesis.md Sections 0 and 9. Do not define lens parameters inline; load from the reference file.


Step 0.5: Fetch Current Price (Mandatory: runs before Stage 1)

Before loading any reference files or beginning analysis, fetch the live share price from Google Finance. This locks in a timestamped reference price used consistently throughout Stage 1, Stage 2, Stage 3, and all outputs.

Price fetch protocol:

  1. Construct the Google Finance URL for the ticker:
    • ASX: https://www.google.com/finance/quote/[TICKER]:ASX
    • NASDAQ: https://www.google.com/finance/quote/[TICKER]:NASDAQ
    • NYSE: https://www.google.com/finance/quote/[TICKER]:NYSE
    • LSE: https://www.google.com/finance/quote/[TICKER]:LON
    • If exchange is unknown, try NASDAQ first, then NYSE.
  2. Use WebFetch on the constructed URL. Extract:
    • Current price (or most recent close if market is closed)
    • Currency
    • Price timestamp (shown on the page: "real-time", "close", or specific time)
    • 52-week high / low if visible
  3. Record and lock as P_current:
    Price source:  Google Finance (google.com/finance)
    Ticker:        [TICKER:EXCHANGE]
    Price:         [X.XX] [Currency]
    As of:         [Timestamp, e.g. "Close 31 Mar 2026, AEST" or "Real-time 14:32 ET"]
    52-week range: [Low] – [High]
  4. If WebFetch fails or returns no parseable price:
    • Try the alternate exchange URL
    • If still unavailable, use WebSearch for the current price and note: Price source: Web search ([source]): Google Finance unavailable for this ticker
    • Never proceed with an assumed or stale price without explicitly flagging it.

P_current is the only price used throughout the analysis. Do not blend it with prices from other sources mid-run. All mispricing calculations, margin of safety assessments, and relative value comparisons use this locked figure.


Step 1: Execute the Analysis

Load and follow the relevant reference files in this order:

  1. Read references/dcf-parameters.md (load DCF methodology before starting Stage 1)
  2. Read references/stage1-equity-analysis.md (execute fully, freeze output)
  3. Read references/stage2-redteam-audit.md (execute using only Stage 1 frozen output; includes Counter-Thesis Construction Section 4 and What Could Go Right Section 4b)
  4. Read references/stage3-thesis-synthesis.md (produce final thesis; includes three-lens verdict table Section 0, Counter-Thesis and What Could Go Right placement Section 0b, and three-lens dedicated sections Section 9)

Stage lock is mandatory. Stage 2 may not introduce new data. Stage 3 may only use what survived Stage 2.


Step 2: Outputs (Standard and Deep Dive)

After completing Stage 3, produce all three outputs:

2a. Chat Summary

Present in conversation:

Three-Lens Verdict Table (mandatory, first element):

LensVerdictBasis
Fundamentals only[BUY / HOLD / SELL / WATCH][One line]
Values-led (mandate)[BUY / HOLD / SELL / AVOID][One line]
Risk-tolerant retail[BUY / HOLD / SELL / CONDITIONAL][One line]

Then:

  • Ticker, price (from Google Finance, P_current), date
  • IV_low / IV_mid / IV_high (post-red-team)
  • Mispricing Score
  • Conviction rating
  • Top 2 surviving risks

Counter-Thesis (named section, 2-3 sentences from Stage 2 COUNTER-THESIS SUMMARY): [Paste verbatim. Counter-thesis strength: X over Y horizon.]

What Could Go Right (named section, 2-3 sentences from Stage 2 WHAT COULD GO RIGHT SUMMARY): [Paste verbatim. Bull case strength: X over Y horizon.]

Final Investment Classification table (mandatory, rendered as a table, all 6 horizons):

HorizonDecisionRationale
1 Month (Reactive)Buy / Hold / Sell[One specific sentence]
3 Months (Tactical)Buy / Hold / Sell[One specific sentence]
6 Months (Transition)Buy / Hold / Sell[One specific sentence]
1 Year (Structural)Buy / Hold / Sell[One specific sentence]
3 Years (Strategic)Buy / Hold / Sell[One specific sentence]
5 Years (Long-Term)Buy / Hold / Sell[One specific sentence]

One closing paragraph (causal, no padding).

2b. Word Document (.docx)

Invoke the docx skill to produce a professional equity research report with this structure:

Cover page:

  • Ticker, company name, date, analyst mode, conviction rating
  • Three-Lens Verdict Table (formatted table, same as chat summary)
  • Methodology note (e.g., "Institutional Sequential Equity Thesis Engine, Standard mode")

Document body (in order):

  1. Executive Summary
  2. Counter-Thesis (named section: 2-3 sentences from Stage 2 COUNTER-THESIS SUMMARY; counter-thesis strength rating)
  3. What Could Go Right (named section: 2-3 sentences from Stage 2 WHAT COULD GO RIGHT SUMMARY; bull case strength rating)
  4. Stage 1: Full Phase 0-6 content
  5. Stage 2: Full adversarial audit content (including Sections 4 and 4b in full)
  6. Stage 3: Full synthesis content including Strategic Decision Matrix and Final Investment Classification
  7. Three-Lens Dedicated Assessment (named section, Lenses A / B / C per stage3-thesis-synthesis.md Section 9)
  8. Sources with timestamps

Formatting notes:

  • Final Investment Classification table rendered as a formatted table with all 6 horizons: not embedded in prose
  • DCF assumptions table
  • Sensitivity table (IV_low / IV_mid / IV_high)
  • Three-Lens Verdict Table on cover page AND at the start of Section 7

2c. Excel Model (.xlsx)

Invoke the xlsx skill to produce a financial model with these tabs:

TabContents
SummaryTicker, price, IV range, mispricing score, three-lens verdict table, conviction
DCF ModelFull two-stage DCF with all inputs in blue, formulas in black (per xlsx skill color standards)
Reverse DCFBack-solved implied growth rate vs. current price
SensitivityWACC x growth rate matrix showing IV per share
FinancialsKey sourced metrics: revenue, FCF, margins, ROE, ROIC, E-C Spread (ROIC−WACC), Reinvestment Rate, Maintenance Capital Burden, Incremental Operating Margin, Cash Conversion Cycle, Sloan Ratio, Interest Coverage, Shareholder Yield, Buyback Execution Alpha, FCF Yield, Altman Z, Rule of 40, Consensus Dispersion
AssumptionsAll inputs with source, date, and deviation notes

Apply xlsx skill color coding: blue = hardcoded inputs, black = formulas, green = cross-sheet links.


Step 3: Quick Scan Output (if Quick Scan mode)

Present in chat only (no Word or Excel):

  • Current price + RSI classification
  • TTM P/E and P/FCF vs. 24-month average (premium/discount %)
  • FCF Yield vs. jurisdiction-appropriate 10Y government bond yield
  • Mispricing Score (0-100) with one-line rationale
  • Simplified three-lens verdict (one line per lens, no dedicated sections):
LensVerdictOne-line basis
Fundamentals only[BUY / HOLD / SELL / WATCH][One line]
Values-led (mandate)[BUY / HOLD / SELL / AVOID][Mandate screen: pass/fail + primary reason]
Risk-tolerant retail[BUY / HOLD / SELL / CONDITIONAL][One line including momentum signal]
  • One sentence: "Worth a deeper look? Yes/No, because [reason]."

Reference Files

Load in order during execution:

  • references/dcf-parameters.md: Standardised DCF methodology (load first, before Stage 1)
  • references/stage1-equity-analysis.md: Full Phase 0-6 instructions
  • references/stage2-redteam-audit.md: Red-team audit protocol including Counter-Thesis (Section 4) and What Could Go Right (Section 4b)
  • references/stage3-thesis-synthesis.md: Integrated thesis construction including three-lens verdict table (Section 0), document placement instructions (Section 0b), and three-lens dedicated sections (Section 9)

Chaining

After delivering output, suggest the most relevant next skill:

  • stock-institutional: institutional investor lens (active weight, tracking error, position sizing, stewardship); chain on request when the user asks for an institutional view, benchmark weight, active weight, or tracking error analysis. State: "Run stock-institutional for the institutional investor perspective."
  • market-research-agent: sector-level context to strengthen the thesis
  • red-team-mode: standalone adversarial challenge of any conclusion
  • exec: compress the thesis into a 1-page investor brief
  • thesis-tracker: formally register this analysis as the founding thesis entry for this position

Chaining note: stock-institutional are planned for a future release of this repository. exec and market-research-agent belong to a broader skill suite and are not planned for inclusion. The investment workflow in this repo is complete without them. Skip any suggestion for a skill you do not have installed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.