- Home
- Skills
- Research & Analysis
- Buy, Hold or Wait: Sourced Long-Term Stock Verdicts
Works with the AI Agents you already use
Buy, Hold or Wait: Sourced Long-Term Stock Verdicts
**What it does** Name a ticker and say whether you already own it.
$7
Buy, Hold or Wait: Sourced Long-Term Stock Verdicts
Example session with this skill installed
Is Costco worth buying for a 10-year hold at today's price, or should I wait?
- Read your context and instructions
- Compiled the buy hold or
Verdict
Verdict: Watchlist
Investor Fit: Good
Time Horizon: 5–10+ years
Confidence: Medium
Main Reason: Costco is one of the best-run retail businesses anywhere, but at $994.61 (47.9x trailing and 43.6x forward earnings) a 10-year buyer's base case works out to only about 6% a year, so the price is the problem, not the business.
Data As Of: Price is the closing price on Oct 20, 2026 ($994.61; a pre-market quote of $988.54 on Oct 21 was not used). The latest periodic report is the Form 10-K for the fiscal year ended Aug 30, 2026, filed Oct 7, 2026. Latest sales are for the September retail month (five weeks ended Oct 4, 2026), released Oct 7, 2026.
You didn't say whether you already own COST, so this verdict is for a new buyer. If you already hold it, the answer is
Hold: the thesis is intact and nothing here justifies selling.
1. What the Company Does
Costco runs membership warehouse clubs: 939 warehouses (647 in the U.S. and Puerto Rico, 115 in Canada, 43 in Mexico, the rest in other international markets), plus e-commerce. It sells a narrow range of products in bulk at very low markups. Members pay an annual fee for access ($65 a year in the U.S.).
Two numbers explain the model
Merchandise is sold close to cost. The gross margin on net sales was 11.09% in FY2026, and SG&A took 9.15% of net sales.
Membership fees carry the profit. Fees were $5.9B in FY2026, about 2% of the $303.2B in total revenue, but equal to about 51% of the $11.7B operating income.
The company had 84.1 million paid members (150.4 million cardholders) and 355,000 employees at fiscal year-end.
2. Business Quality
Great business. Evidence:
Customer stickiness: The renewal rate was 92.3% in the U.S. and Canada and 89.8% worldwide at the end of FY2026, the same as at the end of FY2025 and down from 92.9% and 90.5% at the end of FY2024. A subscription business with renewals that high is rare in retail, though the worldwide rate is now just below 90%.
Members trading up: Executive (higher-tier) members rose to 42.3M from 38.7M (+9.3%), faster than total paid members (+3.8%). They account for 75.6% of sales (Q4 supplemental).
Pricing power: Fees went up in the U.S. and Canada effective Sept 1, 2024. Renewal rates dipped in FY2025, the year the increase took effect (92.9% to 92.3% in the U.S./Canada, 90.5% to 89.8% worldwide), and were flat in FY2026. The company attributes the dip to more memberships sold online, including through digital promotions, which renew at a slightly lower rate. The rate is a trailing measure of renewals 7 to 18 months before the reporting date, so the FY2026 figure already covers renewals made after the increase. Membership fee income still rose about 10% in FY2025 and 11% in FY2026, and the increase supplied about 40% and 30% of that growth.
Scale and a moat that strengthens with growth: More members bring more buying volume, lower prices and more members. FY2026 operating margin edged up to 3.85% of revenue from 3.77%.
Low cyclicality: Food and sundries plus fresh foods were about 53% of FY2026 net sales, and sales rose in every core merchandise category. Separately, the stock's 5-year beta is 0.87; that measures sensitivity to the market, not how cyclical the business is.
- Returns on capital: ROIC is 37.3% and ROE is 28.4% (data-provider calculations).
Weak spots: The margins are very thin, so a small change in pricing or costs moves earnings. The business is capital intensive: $6.4B capex in FY2026, and about $7.5B planned for FY2027.
3. Growth Runway
FY2026 (52 weeks ended Aug 30, 2026):
- Net sales +10.1% to $297.2B, and total revenue +10.1% to $303.2B.
- Comparable sales +8.4% reported and
+6.6% excluding gasoline and FX. Average ticket rose about 5% and shopping frequency about 3%. Higher gas prices added about 130 bps to net sales growth and FX about 49 bps, so headline growth runs ahead of the underlying business.
- Digitally-enabled comparable sales +20.9%.
- Diluted EPS +14.0% to $20.76. Excluding the $0.15 IEEPA tariff-refund benefit (non-recurring, per the company), EPS was $20.61, up about 13.2%. Costco received a further $155M of refunds after year-end and intends to reinvest the majority in member value, so refunds may also affect Q1 FY2027 comparisons.
- Over FY2022–FY2026, revenue grew about 7.5% a year and EPS about 12.1% a year.
Latest data points
Q4 FY2026: Comps +9.4% reported, +6.7% excluding gas and FX. EPS was $6.75, or $6.60 without the tariff benefit, against a reported $6.55 consensus (not verified: the consensus provider is not named, and the source is secondary). Underlying, that was a slight beat, not a big one.
September 2026 (five weeks): Net sales +13.0% to $30.02B. Comps +11.4% reported, +7.6% excluding gas and FX. A later Labor Day added a little over 50 bps.
What is durable: About 3% a year unit growth: 25 net new warehouses in FY2026, and up to 33 openings (5 of them relocations) planned for FY2027, for a planned FY2027 year-end count of 967. Underlying comps in the mid-single digits (+6.6% in FY2026 excluding gas and FX), steady growth in Executive members, and fast-growing digital sales.
What is fading:
- The fee-increase tailwind. Q4 membership fee income grew only 7.3% (7.7% ex-FX), against 11% for the full year.
- Gas inflation, which flatters reported sales but not underlying demand.
Consensus: FY2027 EPS of $22.82 (+10.7% vs. $20.61 underlying; +9.9% vs. $20.76 GAAP) and FY2028 EPS of $25.02 (+9.6%). This is secondary context only.
Growth quality: This is real, organic growth. The 10-K attributes FY2026 sales growth to comparable sales and new warehouses, not acquisitions. There was no dilution (share count was 443.27M at FY2026 year-end, against 443.24M a year earlier), and capex was funded by operating cash flow ($15.83B against $6.44B), although the Japanese subsidiary issued about $0.5B of new notes during the year.
4. Profitability and Balance Sheet
Profitability (FY2026):
Operating income $11.69B (+12.5%). Margin 3.85% of revenue. By segment: U.S. 3.48%, Canada 5.19%, Other International 4.51%.
- Net income $9.23B (+13.9%). Net margin 3.0%. Effective tax rate 24.7%.
Gross margin −3 bps to 11.09% as reported, but +10 bps to 11.22% excluding gasoline price inflation. The 10-K's drivers of that increase: ancillary businesses +19 bps, core merchandise −11 bps (mostly sales mix, as ancillary sales grew faster), and IEEPA refunds net of price reinvestment +3 bps. Core merchandise margin on its own sales rose 15 bps for the year, across all categories, and 18 bps in Q4 (excluding IEEPA). Q4 gross margin was −11 bps reported, +20 bps ex-gas. The reported dip comes from gas prices and sales mix, not from weaker core margins.
- Cash flow: Operating cash flow $15.83B (+18.7%). Capex $6.44B. Free cash flow about $9.39B (+19.8%).
Balance sheet (Aug 30, 2026): Survival risk is negligible.
- Cash and equivalents $20.21B, plus $1.09B short-term investments.
- Total long-term debt $6.18B: $5.0B of parent senior notes plus $1.18B of guaranteed senior notes issued by the Japanese subsidiary (about $0.5B of them issued in FY2026).
- That leaves net cash of about $15.1B before leases, or about $11.1B after subtracting $4.0B of lease liabilities.
- $2.25B of notes mature in fiscal 2027 (May and June 2027). Cash alone covers that about 9 times.
- Interest expense was $145M, against operating income about 81 times larger.
- No short-term borrowings at year-end.
Dilution: None of note. The share count is flat year over year. Buybacks were $848M in FY2026 (891,000 shares at an average of $950.90), and $1.115B of authorization remains, expiring January 2027.
Dividend: The quarterly dividend is $1.47 (raised from $1.30 in April 2026). Dividends paid were $2.46B, about 26% of free cash flow.
5. Valuation
Current multiples (price $994.61, Oct 20, 2026 close; market cap about $440.7B):
| Measure | Value |
|---|---|
| Trailing P/E (GAAP EPS $20.76) | 47.9x |
| Trailing P/E ex tariff benefit (EPS $20.61) | 48.3x |
| Forward P/E (FY2027 consensus EPS $22.82) | 43.6x |
| P/E on FY2028 consensus EPS ($25.02) | 39.8x |
| Free cash flow yield (FCF $9.39B / market cap) | 2.1% |
| EV/EBITDA | 30.3x |
| PEG ratio | 4.0 |
| Dividend yield | 0.59% |
| Gross buyback yield (FY2026 repurchases / market cap) | about 0.2% |
Against its own history:
Last five fiscal years: Quarter-end trailing P/E ranged from 36.7x (Feb 2023) to 62.4x (Feb 2025) and averaged about 46.6x. A second provider puts the 5-year average at 45.4x. On that window the stock is about average.
Longer view: Calendar year-end P/E was about 24.8x in 2016, 26.4x in 2017, 24.8x in 2018 and 36.5x in 2020 (that provider's method reads today at 50.0x, so compare the trend, not the decimals).
Takeaway: On that provider's figures, the year-end multiple roughly doubled, from about 25x (2016–2018) to about 46–54x (2023–2025), so part of the past decade's return came from re-rating. A buyer today cannot count on that happening again.
Against peers (Oct 20, 2026 close):
| Company | Trailing P/E | Forward P/E |
|---|---|---|
| Costco | 47.9x | 43.6x |
| Walmart | 42.7x | 39.2x |
| BJ's Wholesale | 22.4x | 21.1x |
Costco trades at a premium even to Walmart, and at about twice its closest club peer.
What is priced in: a 10-year model. It starts from $20.61 underlying EPS, pays dividends at a 27% payout (FY2026 was 26.6%), excludes taxes and special dividends, and assumes you sell at the end of year 10.
| Case | EPS growth / yr | P/E in year 10 | Year-10 EPS | Year-10 price | Annual return at $994.61 |
|---|---|---|---|---|---|
| Bear | 6% | 28x | $36.91 | ~$1,033 | ~1.2% |
| Base | 9% | 35x | $48.79 | ~$1,708 | ~6.3% |
| Bull | 11% | 42x | $58.52 | ~$2,458 | ~10.1% |
Why these assumptions
Base-case growth of 9% sits just below consensus for the next two years (10.7%, then 9.6%) and below the FY2022–26 EPS pace of 12.1%. The fee-increase boost is fading, and unit growth is about 3% a year.
Base exit P/E of 35x is slightly below the lowest quarter-end multiple of the past five years. My assumption is that the multiple drifts down as growth slows.
The other cases: The bear exit of 28x is a little above the 2016–2018 year-end range (24.8–26.4x). The bull exit of 42x is about today's forward multiple.
Sensitivity
- At 9% growth, exits of 25x / 30x / 35x / 40x / 45x give about 2.9% / 4.7% / 6.3% / 7.6% / 8.9% a year.
- At a 35x exit, growth of 7% / 8% / 9% / 10% / 11% / 12% gives about 4.3% / 5.3% / 6.3% / 7.2% / 8.2% / 9.2% a year.
What the current price requires: To earn 10% a year from $994.61, EPS must compound about
12.8% a year for ten years with a 35x exit, or about
10.1% a year with the market still paying 45x in 2036. Either today's growth lasts a decade and the premium multiple survives, or you accept mid-single-digit returns.
Verdict on valuation:
Expensive, but not absurd. Today's price is fair if you assume the 5-year average multiple lasts forever, and expensive against any longer history or the peers. On the base case, there is no margin of safety at the current price.
6. Is Now a Good Time to Buy?
Entry Timing: Wait for pullback. Once the price is in the Fair Zone, buy slowly and dollar-cost average.
Buy Zone: Attractive below about
$710. That is about 31x FY2027 consensus EPS, where the base case returns 10% or more a year.
Fair Zone: About
$710–$930, or about 31–41x FY2027 EPS. In this zone the base case returns 7–10% a year (about 8% a year at $850).
Expensive Zone: Above about
$930, where the base case falls below 7% a year. Today's $994.61 is about 7% above the top of the Fair Zone.
These bands rest on the base case in section 5 and on two hurdles I chose: 10% a year for "attractive" and 7% a year for "fair". Change either and the bands move. They are research estimates, not price targets.
Reconsider Thesis If: Any of the following happens:
- The U.S./Canada renewal rate falls below 90% and stays there (it is 92.3% now, down from 92.9% at the end of FY2024). The worldwide rate is already 89.8%, but the company says it is held down by membership growth in some international markets and by online sign-ups, so the U.S./Canada rate, in the core market with 762 of the 939 warehouses, is the cleaner test of the model.
- Paid membership growth stalls (it is +3.8% now).
- Comparable sales excluding gas and FX run below about 3% for several quarters (they were +6.6% in FY2026).
- Core merchandise margin on its own sales starts declining (it rose 15 bps in FY2026) without a matching gain in fees.
Best Strategy: I would not build a full position at $995. I would wait for the Fair Zone (about $930 or lower), accumulate there in several tranches, and treat the Buy Zone as the point where the risk/reward becomes clearly attractive.
The Fair Zone is not hypothetical: COST closed at $923.52 on Oct 5 and $896.48 on Sep 24, 2026. The cost of being impatient is modest over ten years: in the base case, buying at $995 instead of $930 lowers the expected return from about 7.0% to about 6.3% a year. The bigger risk is the exit multiple, not the entry price.
Portfolio Fit: Without your full portfolio, I can only judge this as a standalone investment. For reference, this framework's generic ranges are: starter position 0.5–2%, medium 2–5%, core 5–10% maximum for a high-conviction diversified investor. These are the framework's reference ranges, not a recommendation for your money.
Timing context:
- The stock is up about 11% since the close before Q4 results ($896.48 on Sep 24). It rose 2.9% the day after the report.
- It sits about 6% above its 50-day average ($935.58) and about 3% above its 200-day average ($967.49).
- RSI is 73, which is short-term stretched (secondary evidence only).
- It is 9.3% below the May 19, 2026 high of $1,096.50 and 17.8% above the Dec 16, 2025 low of $844.06.
- The next earnings report is expected around Dec 10, 2026 (data-provider estimate, not verified with the company).
- The stock goes ex-dividend on Oct 30, 2026.
7. Key Risks
Fundamental
- The fee-increase tailwind is fading: Q4 membership income growth slowed to 7.3%, against 11% for the year. Fees are about half of operating income, so slower fee growth matters directly.
- Reported gross margin is diluted by gasoline prices and by faster growth in ancillary businesses (−3 bps for the year and −11 bps in Q4 as reported, but up excluding gas). Core merchandise margin on its own sales is still rising; a turn lower, for example from heavier price reinvestment, would be the real warning.
- Competition from Walmart/Sam's Club, BJ's and online retailers. Competition from other retailers and warehouse club operators is a listed 10-K risk factor.
Valuation
- This is the main risk. At about 48x trailing earnings, a de-rating to the roughly 25–37x seen at the 2016–2018 and 2020 year-ends could take most of a decade's earnings growth: the bear case is about 1% a year.
- The stock is priced for steady execution. A single soft quarter on comps or renewals could hit the multiple hard.
Financial
- Low. The company has net cash, and its $2.25B of fiscal 2027 maturities are covered about 9 times by cash.
- Capex is rising to about $7.5B, close to half of FY2026 operating cash flow.
Management/Governance
- No red flags found.
- Insider sales since Q4 look small: three officer Form 144 notices of proposed sales totalling 4,101 shares (about $3.8M), all from vested restricted stock. Form 4 filings were not checked.
- The FY2027 executive bonus plan ties pay to sales, pre-tax income, and environmental and social metrics, with the sales and pre-tax income goals paying 0–120% of target.
Macro/Sector
- Tariffs. The 10-K says the company paid about $500M in IEEPA tariffs before the Supreme Court ruled them unlawful in February 2026. It received $184M during FY2026 ($174M of refunds plus $10M of interest) and a further $155M after year-end, and intends to reinvest the majority in member value.
- Consumer class actions seek pass-through of those refunds. A motion to dismiss was denied in July 2026, and a motion for discretionary appellate review of that ruling is pending.
- Other pending litigation: wage-and-hour claims, a certified employee class action over a second-job restriction (trial set for March 2027), website-tracker privacy suits with a state attorney general inquiry, product-labeling claims, email-marketing class actions (one being settled, one filed in July 2026), and remaining opioid matters. The company says none is expected to be material in aggregate.
- Gas prices and FX swing reported sales in both directions.
Thesis-Breaking
- A sustained fall in renewal rates or paid members. That would mean the membership model itself is weakening, and the fee stream is what supports the valuation.
Red Flags: Two checklist items are present but judged minor, and a third shows up only in reported figures. None is serious enough to downgrade the verdict, which is already Watchlist on valuation.
Present, judged minor: (1) Regulatory investigations: a state attorney general has issued civil investigative demands about website trackers. The company is responding and does not expect pending matters to be material. (2) Revenue growth slowing while valuation remains high: total revenue is not slowing (+10.1% in FY2026, September net sales +13.0%), but membership fee growth, the source of about half of operating income, slowed to 7.3% in Q4 against 11% for the year, with the stock near 48x.
Present in reported figures only: Gross margin compression: reported gross margin fell 3 bps for the year and 11 bps in Q4, but rose 10 bps and 20 bps excluding gas, and core merchandise margin on its own sales rose (section 4). Not treated as a flag.
Checked and clear: negative FCF with high debt; constant dilution; insider enrichment; weak or confusing disclosures (the 10-K and Q4 supplemental break out gas, FX, and core versus ancillary margin effects); customer or single-product concentration; dependence on one commodity cycle (gas swings reported sales, but core merchandise is about 79% of net sales); declining return on capital (ROIC rose to 37% from about 31% in FY2022–23); repeated guidance misses (Costco gives no numeric guidance); "adjusted EBITDA" games (results are GAAP); high short interest (1.7% of float).
Not checked: auditor concerns (the auditor's report was not read); management credibility beyond the absence of guidance misses (the earnings-call transcript was not reviewed); stock promotion behavior. Related-party transactions and Form 4 insider filings were not checked either.
8. Long-Term Holder Guidance
I would be comfortable holding if: The U.S./Canada renewal rate stays around 90% or higher (92.3% now; the worldwide rate, 89.8%, runs lower for the reasons in section 6), paid members keep growing (+3.8%), comps excluding gas and FX stay mid-single-digit (+6.6% in FY2026), and operating margin holds near 3.8–3.9%.
I would add only if: The price drops into the Fair Zone (about $930 or lower, about 41x FY2027 EPS) with those metrics intact. I would add aggressively only in the Buy Zone (about $710 or lower). The other route is EPS compounding visibly above about 11% a year, which would move the zones up.
I would stop adding if: The price runs well above about $930 without matching earnings upgrades, or the trailing P/E climbs back toward the 53–62x seen in 2024–25.
I would reconsider the thesis if: Any of these happens:
- The U.S./Canada renewal rate drops below 90% for several quarters.
- Paid member growth goes flat or negative.
- Comps excluding gas and FX fall below about 3% for several quarters.
- Core merchandise margin on its own sales starts declining without a matching gain in fees.
I would seriously consider selling only if: Any of these happens:
- The membership model breaks, shown by a collapse in renewals.
- An accounting or governance problem emerges.
- Management turns to debt-funded buybacks or large acquisitions that damage the balance sheet.
- The position grows into a dangerously large share of your portfolio.
9. Score
Business Quality: 19/20
Growth Runway: 11/15
Profitability: 12/15
Balance Sheet: 14/15
Valuation: 5/15
Management & Capital Allocation: 9/10
Timing / Entry Point: 4/10
Total: 74/100 (65–74: Mixed; suitable only with caution)
On the five business categories alone (everything except Valuation and Timing), it scores 65/75, about 87%. Valuation and timing pull the total into the "caution" band.
Final Takeaway
Costco is a great company, but at $994.61 it is not a great stock to buy today. The business is very strong: U.S./Canada renewals at 92.3%, unit growth of about 3% a year, net cash, and EPS up 13% underlying. But at about 48x trailing and 44x forward earnings, earning 10% a year from here would take about 12.8% EPS growth for a decade with a 35x exit, or about 10.1% with the market still paying 45x. On reasonable assumptions, a 10-year buyer from here earns about 6% a year, about 1% in the bear case, and only about 10% if everything goes right.
For a new 10-year buyer, my answer is
wait. I would keep it on the watchlist, start building gradually only below about $930 (it traded there as recently as Oct 5), and see the risk/reward as clearly attractive below about $710. If you already own it, hold: nothing here is a reason to sell.
Sources
- Costco Q4 and fiscal 2026 earnings release (Exhibit 99.1 to Form 8-K), dated and filed Sep 24, 2026, SEC EDGAR.
- Costco Q4 FY2026 Supplemental Information (Exhibit 99.2 to the same Form 8-K), Sep 24, 2026, SEC EDGAR.
- Costco Form 10-K for the fiscal year ended Aug 30, 2026, filed Oct 7, 2026, SEC EDGAR. Includes the XBRL notes on commitments and contingencies and on segment reporting (including net sales by merchandise category). This is the latest periodic report: after a fiscal fourth quarter there is no 10-Q.
- Costco Form 10-K for the fiscal year ended Aug 31, 2025, filed Oct 8, 2025, SEC EDGAR (prior-year renewal rate, member counts and membership fees, and the company's explanation of the FY2025 renewal dip).
- Costco Form 10-K for the fiscal year ended Sep 1, 2024, filed Oct 9, 2024, SEC EDGAR (end-FY2024 renewal rates).
- Costco September sales results release, Oct 7, 2026 (GlobeNewswire, read via Finviz).
- Costco Form 8-K, Oct 14, 2026 (quarterly dividend, Exhibit 99.1), and Form 8-K, Oct 16, 2026 (Item 5.02, FY2027 executive bonus plan), SEC EDGAR.
- SEC Form 144 notices filed Sep 29, 2026 (two) and Oct 13, 2026; Costco EDGAR filing index viewed Oct 21, 2026.
- StockAnalysis.com: COST quote, statistics, annual and quarterly ratios, financials, forecast (consensus updated Oct 15, 2026) and dividend pages; WMT and BJ statistics. All at the Oct 20, 2026 close, viewed Oct 21, 2026.
- Yahoo Finance COST historical prices (Sep 15 to Oct 20, 2026), viewed Oct 21, 2026.
- StockScan COST price history (52-week high and low dates), viewed Oct 21, 2026.
- Bullfincher COST P/E (5-year average quarterly P/E), as of Oct 20, 2026.
- CompaniesMarketCap COST P/E history (calendar year-end P/E), viewed Oct 21, 2026.
- Tradingpedia article on the Q4 results (URL dated Sep 25, 2026; no date in the article text): Q4 consensus EPS of $6.55; secondary source.
Not verified
- The Q4 consensus estimate (the provider is not named).
- The Dec 10, 2026 earnings date (a data-provider estimate, not announced by Costco).
- The 10-year average P/E. GuruFocus, FullRatio and Macrotrends refused the request, so calendar year-end values from one provider are used instead, and its method differs from StockAnalysis's.
- ROIC, ROE, EV/EBITDA and PEG are the data provider's calculations, not company-reported figures.
- No earnings-call transcript was reviewed.
- Form 4 insider filings, the auditor's report and related-party transactions were not reviewed.
This is investment research, not personalized financial advice. I do not know your full financial situation, tax status, risk tolerance, or portfolio concentration.
Connects securely to your tools. The creator never sees your data.
About this skill
What it does
Name a ticker and say whether you already own it. Before giving any verdict, the agent fetches the current price with its date, the latest annual and quarterly reports (confirmed as the latest on the company's investor relations page or SEC EDGAR), the latest earnings release or call, and any material news and 8-Ks since. For an ETF it fetches the fund's fact sheet or prospectus, expense ratio, holdings, sector and country weights, and tracking instead, and scores the fund on its own rubric. If you do not say whether you own the security, it asks once, then answers for a new buyer and adds one line for current holders. It then works through eight sections: business quality, growth runway, profitability, balance sheet and survival risk, valuation, timing and entry point, risks, and long-term holder guidance. The answer opens with a six-line verdict block (verdict, investor fit, time horizon, confidence, main reason, data as-of date) and ends with a 100-point score, a plain-English takeaway, a dated source list, and a list of the figures it could not verify.
Why this instead of a prompt
A plain "should I buy X?" prompt can answer from training data: last year's price, an old P/E, guidance that has since changed. This skill makes fresh data mandatory. It lists the figures the agent must never take from memory (price, market cap, multiples, growth, guidance, balance sheet, insider transactions, dividend yield, earnings date and more), requires a source and as-of date for every figure that drives the verdict, and tells the agent to mark a figure unverified when it cannot be fetched or sources disagree.
It is also written for a specific investor: long-only, 3 to 10+ year horizon, dislikes selling. For an existing holding, Hold is the baseline. Sell or Trim is only allowed when one of ten listed thesis-broken conditions applies (for example fraud risk, severe dilution, solvency risk, or a dangerously concentrated position). And it separates a great company from a great stock: a high-quality business does not get a Buy if the valuation is extreme. In the demo, Costco scores 19/20 on business quality but 5/15 on valuation, so the verdict for a new buyer is Watchlist, with price zones for when that changes, and Hold for anyone who already owns it.
Key features
- Mandatory Freshness Rule: current price with date, latest 10-K/10-Q (or 20-F/6-K or local equivalent), latest earnings release or call, and 8-Ks since the last report, all fetched before the verdict. ETFs get their own fetch list. If SEC EDGAR blocks the agent, it reads the same filing from the company's IR site or through its fetch tool, and after a fiscal fourth quarter it treats the 10-K as the latest periodic report instead of hunting for a 10-Q.
- Eight decision labels (Strong Buy, Buy, Starter Buy, Add Slowly, Hold, Watchlist, Avoid, Sell or Trim) with written criteria for each. Watchlist always comes with a condition for reconsidering.
- Valuation by company type: P/E, forward P/E, PEG, EV/EBITDA and FCF yield for profitable companies; EV/Sales, Rule of 40 and dilution-adjusted valuation for growth companies; price/book, NAV and dividend sustainability for asset-heavy ones. Always answers "what growth is already priced in?" and whether there is a margin of safety.
- Seven entry-timing categories (from "Attractive now" to "Avoid regardless of price") plus Buy, Fair and Expensive zones. Price targets must show their assumptions.
- Risks split six ways: fundamental, valuation, financial, management/governance, macro/sector, and thesis-breaking.
- A mandatory long-term holder section: comfortable holding if, add only if, stop adding if, reconsider the thesis if, seriously consider selling only if.
- A 100-point score across seven categories with interpretation bands (a six-category variant for ETFs: cost, diversification, holdings quality, valuation of holdings, tracking and liquidity, overlap), and a 16-item red-flag checklist that downgrades the verdict when several flags appear. The report states each flag found and names any checklist item it did not check.
- A separate section layout for ETFs, plus extra checklists for high-growth tech, AI stocks, banks, REITs and Chinese ADRs / foreign stocks.
- Portfolio fit (sector and geographic concentration, overlap) when you give portfolio context; otherwise it states that it is judging the security on its own. It gives generic position-size reference ranges, never a size for your money.
- A fixed answer template, so every report has the same sections in the same order, including fixed slots for portfolio fit and the red-flag result.
- Hard rule: it never places, modifies or cancels orders and never logs in to a brokerage account, even if a tool would allow it.
Who it is for
Individual long-term investors who want a structured, sourced second opinion before buying or adding to a position; holders who want to know whether a stock is still a hold, worth adding to, or showing real thesis-broken signals; and students learning fundamental analysis who want a repeatable framework. It is not for short-term trading, options, or short selling.
How it is built
A single SKILL.md (809 lines of plain Markdown instructions). No scripts, no dependencies, no bundled data, no API keys. It uses whatever web search and web fetch tools the agent already has, or a market-data tool if one is connected. The only network traffic is the agent's own read-only requests to public sources such as company IR pages, SEC EDGAR, financial data sites and news.
Known limitations
- The result is only as current as the agent's web access. Without browsing, the skill says so, falls back to a framework analysis using figures you provide, labels them user-provided or unverified, and lowers Confidence.
- Free sources have gaps, and the report lists them rather than filling them in. In the demo, three data sites refused the request, so the 10-year average P/E is marked not verified and the longer valuation history rests on one provider's year-end snapshots. The peer comparison is trailing and forward P/E against two peers, with no EV/EBITDA or FCF-yield comparison. No earnings-call transcript was read, and the auditor's report, related-party transactions and Form 4 insider filings were not checked.
- Scores are analyst judgment inside a fixed rubric, not a quantitative model. There is no fixed mapping from score to verdict, and Timing overlaps with Valuation, so an expensive stock loses points in both.
- The skill has no explicit rule for GAAP versus adjusted EPS or for stripping one-time items, and no industry checklist for consumer staples or retail. The agent decides these case by case (the demo shows both GAAP EPS and EPS excluding a tariff-refund benefit).
- When it cannot ask whether you own the security, the current-holder answer is one line (in the demo, "Hold" with a one-sentence reason); the detailed holder thresholds are in the long-term holder section.
- The demo covers a single stock. No ETF run is shown, so the ETF fetch list, score variant and layout have not been exercised in a demo.
- One security per run. No screening, backtesting, alerts or portfolio optimization.
- Investment research, not personalized financial advice. It gives generic position-size reference ranges, never a size for your money, and does not account for your taxes or income.
- The demo below is one real run. It was checked against its sources afterwards and several statements were corrected before posting (margin bridge, renewal trend, litigation, debt); check key figures yourself before acting on any report.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.
This is investment research on one security at a time, not personalized financial advice: it gives only generic position-size reference ranges, never a size for your money, and it does not consider your taxes, screen for stocks, backtest, or place trades. Its figures are only as current as your agent's web access; without browsing it says so, works from figures you provide, labels them unverified and lowers its confidence. Free sources have gaps, and the report lists what it could not get instead of filling it in: in the demo, the 10-year average P/E is marked not verified, the peer comparison is P/E only against two peers, and no earnings-call transcript, auditor's report or Form 4 filing was reviewed. When it cannot ask whether you own the security, the current-holder answer is a single line, with the detail in the holder section. The demo shows a stock; the ETF branch has no demo run. The 100-point score is structured judgment, not a quantitative model, and the skill leaves some calls to the agent, such as GAAP versus adjusted EPS, how to treat one-time items, and which industry metrics matter for sectors it has no checklist for (for example consumer staples and retail). The demo below is one real run. It was checked against its sources afterwards and several statements were corrected before posting (margin bridge, renewal trend, litigation, debt); check key figures yourself before acting on any report.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 27 days ago
Requires
Nothing to install.
Permissions
the skill's core rule is that the agent must not judge a stock from memory. It has to look up today's price and the company's latest reports before it gives a verdict, so it needs to search the web and read public pages and filings. It only reads; it does not log in anywhere, does not send your data to any service, and never connects to a brokerage or places, changes or cancels orders. File reading and a shell are optional and only used to read documents, such as a filing you provide or a PDF the agent downloaded.