Guide · Agent skill
How to use the Investor Lenses agent skill: setup and examples
Updated · written with AI from the project's own documentation and checked against our catalogue data
Investor Lenses is an agent skill that makes an AI agent judge a company or investment through the reasoning style of named investors, such as Graham, Buffett, Munger and Fisher. It works on evidence you have already gathered and does not collect data itself. After this guide you will know what an agent skill is, how to install this one, how to ask for a lens by describing your situation, and how to read the results without treating them as investment advice.
Investor LensesHKUDS/Vibe-Trading · MIT Download the skill Quick start
- 1Download the ZIP. Download the investor-lenses ZIP from this page. Keep its contents together.
- 2Unzip the skill. Unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill there instead.
- 4Write down your evidence. Collect the financials, prices, disclosures and timeline you already have before choosing a lens.
- 5Pick a lens by situation. Use the situation table in SKILL.md to choose a starting lens. Do not choose by the verdict you hope for.
- 6Run the lens and check its veto rules. Ask the agent to walk the priority signals in order and stop if a hard disqualifier applies.
- 7Stack and compare. Add one or two lenses that can disagree with the first, run each independently, and report each disagreement's root cause.
What is an agent skill, and what does investor-lenses do?
An agent skill is a folder of instructions that an AI agent can load when a task matches it. This one is named investor-lenses and is in the analysis category. Its main file is SKILL.md, and it includes 12 reference files, one per lens.
A lens is a fixed, ordered set of questions plus explicit veto rules, based on one investor's publicly described methodology. Each lens ranks certain signals first, lists hard disqualifiers that end the run, and documents the conditions where it tends to fail. The skill's authors describe each lens as their own reconstruction, not something the named investor wrote or endorsed.
- Graham: deep value and balance sheet
- Buffett: quality franchise and owner earnings
- Munger: inversion, looking for how a plan could fail
- Fisher: scuttlebutt growth
- Lynch: growth at a reasonable price (GARP)
- Marks: cycle positioning
- Dalio: debt cycle and portfolio shape
- Chanos: forensic short
- Bogle: cost drag and mean reversion
- Duan Yongping: right business, right people
- Feng Liu: weak-system contrarian
- Qiu Guolu: three questions on industry structure
How do you install the investor-lenses skill?
Download the ZIP of the investor-lenses skill from this page and unzip it. You can install it in one of two ways. For a personal setup, place the skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Keep the folder intact so SKILL.md and the references folder stay together.
- Use the folder for your personal setup (~/.claude/skills/) or for one project (.claude/skills/) depending on where you want it available.
- The Claude apps route is an upload through Settings → Capabilities → Skills.
How do you trigger the skill with a natural request?
You do not need a special command. Describe the question you want answered, and say which kind of judgement you need. The skill's selection guide maps common situations to lenses. For example, a question about whether a cheap, asset-heavy stock is cheap enough points toward Graham. A question about whether the story is real points toward Fisher. A question about how the business could fail points toward Munger.
Give the agent your evidence first, or say clearly what it is. The skill is designed to run on evidence already collected, so a request that includes financials, prices, filings and a timeline will work better than a vague question.
- Example: 'Run the Graham lens on this company using the balance sheet and price history I pasted.'
- Example: 'Pre-mortem this decision with Munger inversion. Here is my research memo.'
- Example: 'Compare the Buffett and Chanos lenses on these filings and tell me where they disagree.'
- Example: 'The accounting here looks odd. Apply the Chanos lens to these statements.'
What does a good run look like?
The skill sets out a five-step run. First, freeze the evidence, meaning you write down what is known before choosing a lens. Second, choose the lens by the situation, not by the answer you want. Third, walk the priority signals in the order the lens sets. Fourth, if a hard disqualifier applies, stop and report the veto. Fifth, check whether the case falls into one of the lens's documented failure regimes, then report the result.
When you stack lenses, the skill recommends two or three at most. It also says to run each lens independently, classify each disagreement by root cause (horizon, caliber, risk preference or information), and never average the verdicts into a single score.
- Freeze the evidence before choosing a lens.
- Stack two or three lenses that can actually disagree.
- Classify each disagreement as horizon, caliber, risk preference or information.
- Report the losing lens's strongest point in its own terms.
What does the skill need, and what are the safety notes?
The skill lists no required dependencies. It does not run scripts and does not use the network, according to the skill's listing. This means it works only with the text you give it and the reference files in its folder. It does not fetch prices or filings for you.
The skill frames every output as an analytical exercise, not investment advice. Each lens verdict holds only under that lens. Short-side lenses carry risk: the Chanos reference notes that short exposure carries unbounded loss.
- Requirements: none listed.
- Scripts: none.
- Network access: none.
- Output is analysis, not a buy or sell recommendation.
- Do not pick the lens that gives the answer you want.
What tips make the results more useful?
Match the lens to the situation. Each reference file lists the markets and situations where its lens is sharp and where it is blind. Buffett is weak on fast-changing tech, for example, and Qiu Guolu is weak on early-stage industries. Check the lens's failure modes before you rely on its verdict.
Use a second lens to find the assumption your decision rests on. If two lenses conflict, the disagreement shows which fact or time horizon matters most. Do not treat a 2-to-1 split as a vote.
- Read the lens file's market-fit note before running it.
- Write your evidence down first so the lens is not applied to data collected afterward.
- Ask the agent to state whether your case falls into a documented failure regime.
- Ask for the strongest point of the losing lens, not just the winner.
What people use it for
- Comparing how Graham and Buffett judge the same company on the same financials.
- Running a Munger inversion pre-mortem on a decision you have already researched.
- Checking whether a company's accounting matches its cash economics with the Chanos lens.
- Testing whether a growth stock's valuation is reasonable with the Lynch GARP lens.
- Studying investment methodology by running one repeatable procedure across several cases.
Frequently asked questions
Is the investor-lenses skill free to use?
The skill is published under the MIT licence, as listed on its page. Check the licence file in the repository for the full terms before you reuse or redistribute it.
What do I need to install investor-lenses?
The skill lists no required dependencies. You need an agent setup that supports skills, and you place the unzipped folder in ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills in the Claude apps.
Does investor-lenses fetch stock data or run scripts?
No. The skill runs no scripts and uses no network. It applies its lenses to evidence you supply and does not gather data itself.
Will investor-lenses tell me whether to buy or sell a stock?
No. The skill states its outputs are analytical exercises, not investment advice. Each lens verdict is conditional on that lens and does not mean an investment is good or bad.
How is this different from just asking an AI to analyse a stock?
The skill fixes the procedure. Each lens has an ordered checklist, hard disqualifiers and documented failure modes, so the same method runs the same way each time. It also asks you to freeze your evidence first and to report disagreements between lenses instead of averaging them.
Why doesn't my agent use the skill when I ask?
The skill can only be used if it is installed in the location your setup reads from. Check that the folder is unzipped in ~/.claude/skills/ or the project's .claude/skills/, or that it was uploaded under Settings → Capabilities → Skills, and that the skill folder is intact. Then name the situation or the lens you want in your request.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use investor-lenses if you want to compare how different value and growth frameworks judge the same evidence, or if you want a structured pre-mortem or second opinion on a decision you have already researched. Pick something else if you want a buy or sell recommendation, or if you plan to run only the lens that gives the answer you want.
Download the skill