{"slug": "investment-skills-lexicon-how-great-investors-think", "title": "Investment Skills Lexicon: How great investors think", "summary": "Finterm AI released an open-source Investment Skills Lexicon that distills Warren Buffett's investment framework from 72 primary documents (950,000 words) into 65 evidence-bounded decision cards for use by AI agents. The project processed roughly 3.4 billion tokens to extract 3,809 insights, enabling agents to apply Buffett's mental models to financial analysis without loading the full corpus. The framework supports three workflows—financial analysis, memo review, and thesis evaluation—and is available via the Agent Skills standard for tools like Claude Code and Codex.", "body_md": "**How great investors think, distilled from primary sources.**\n\nThe primary Buffett and Berkshire corpus runs 72 documents and about 950,000 words. This project extracts, analyzes, and condenses it (3,809 extracted insights and roughly 3.4 billion tokens of processing) into 65 decision cards that an agent loads 5–12 at a time.\n\nYou can always ask an agent to “analyze this like Warren Buffett.” But the goal is not to role play an investor’s voice. Rather, we want an agent to apply the underlying frameworks and mental models used by that investor to real filings, memos, and theses without searching or reading the whole corpus into context.\n\nThe idea of this project is to spend the tokens and do heavy thinking once, then reuse the insights in a context-efficient way.\n\nThis works with Claude Code, Codex, or any agent that reads the open\n[Agent Skills](https://agentskills.io) standard.\n\n**The Buffett investment framework** distills ~2,300 pages of primary Buffett and\nBerkshire writing into **65 evidence-bounded decision cards** an agent can use.\n\nIt is an agent skill for **evidence-bounded financial analysis, investment-memo review,\nand thesis evaluation**. It offers **three focused workflows** that surface assumptions,\ncounterarguments, missing evidence, and invalidation conditions.\n\nThe framework helps an agent reason about a business.\nIt does not issue buy, sell, hold, entry-price, position-size, or trade instructions.\nThe complete agent instructions are in\n[SKILL.md](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/SKILL.md), which is self-contained once\ninstalled.\n\n| Stage | Contents | Scale |\n|---|---|---|\n| Source corpus | 48 Berkshire shareholder letters (1977–2024), Greg Abel’s 2025 transition letter, the 1957–1970 Buffett Partnership letter compilation, other Buffett and Berkshire writings, The Essays of Warren Buffett, and Buffett: The Making of an American Capitalist |\n72 documents (2,300 pages, 950,000 words) |\n| Extraction | Source-native insights distilled from the corpus | 3,809 insights |\n| Processing | Extraction and synthesis ran across Codex Sol and Claude Fable | about 3.4 billion tokens |\n| Skill lexicon | Final task-facing output | 65 cards · 8 modules · about 51 pages |\n\nThis “lexicon” is a growing collection of insights. Decades of investment thinking sit in writings, interviews, and filings that few people or even agents have time to analyze deeply. So we’re extracting these piece by piece.\n\nThe first piece is the Buffett framework. Queued next: Munger, Graham, Marks, Fisher, Lynch, Sleep and Zakaria, Grantham, Greenblatt, and Druckenmiller.\n\nCheck back here as we’re adding a new one each week!\nPrimary sources and framework focus for each are in the [roadmap](/finterm-ai/investment-skills/blob/main/docs/roadmap.md).\n\nInstall with the [ skills CLI](https://github.com/vercel-labs/skills) (more options\nunder\n\n[Other Install Methods](#other-install-methods)):\n\n```\nnpx skills add finterm-ai/investment-skills --skill buffett-investment-framework --yes\n```\n\nThen ask in your own words and attach the evidence. The skill picks the workflow and loads only the cards that fit:\n\n```\nWhat are the real owner earnings in these filings?\nReview this investment memo and tell me which claims hold up.\nHere is my thesis on this company. What would prove it wrong?\n```\n\nMost agents load the skill from its description, so naming it is optional.\nIn Codex, type `$`\n\nto mention it explicitly.\n\n| Workflow | Starting load | What it returns |\n|---|---|---|\n| Financial analysis | `F01` , `F02` , `F05` , `F06` , `F07` |\nFiling inventory, reported-to-owner bridge, normalized segments, returns, obligations, assumptions, and blocked calculations |\n| Memo review | `D01` , `D03` , `B02` , `V01` , `V05` , `V06` , `R01` , `R07` |\nOne dispositioned row per material claim, with counterevidence and missing evidence |\n| Thesis evaluation | `D02` , `D03` , `B01` , `B06` , `V01` , `V04` , `R01` , `R07` |\nComponent map, mechanism and valuation tests, owner-harm paths, and invalidation conditions |\n\nThe router starts with 5–8 cards, adds only material management, allocation, financing, or specialized overlays. It caps at 12 cards in one pass to preserve context and stay focused.\n\n| Module | Cards | Decision use |\n|---|---|---|\n|\n\n`D01`\n\n–`D09`\n\n[Business economics](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/02-business-economics.md)`B01`\n\n–`B07`\n\n[Management and governance](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/03-management-governance.md)`M01`\n\n–`M08`\n\n[Financial reality](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/04-financial-reality.md)`F01`\n\n–`F07`\n\n[Valuation](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/05-valuation.md)`V01`\n\n–`V06`\n\n[Capital allocation](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/06-capital-allocation.md)`C01`\n\n–`C09`\n\n[Risk and monitoring](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/07-risk-monitoring.md)`R01`\n\n–`R07`\n\n[Specialized overlays](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/08-specialized-overlays.md)`S01`\n\n–`S12`\n\nEvery card uses the same contract: decision question, guidance, use condition, analytical actions, observable output, limits, readable source basis, and abbreviated corroboration citations.\n\nEvery completed analysis reports:\n\n- Question, horizon, scope, and exclusions.\n- Evidence received and material missing inputs.\n- The 5–12 cards loaded and why.\n- Sourced calculations, bridges, and mechanism tests.\n- Supported, challenged, or unresolved findings.\n- Counterevidence and alternate mechanisms.\n- Limits and blocked branches.\n- Monitoring evidence and invalidation conditions.\n\nThe result ends with an analytical summary.\n\nThe framework is an **editorial synthesis of published Buffett and Berkshire writings**,\nnot a transcription or an attempt to imitate Buffett’s voice.\nThe development corpus drew primarily from Berkshire Hathaway annual letters, Buffett’s\n2015 50th-anniversary essay, and “The Superinvestors of Graham-and-Doddsville.”\n\nThe material was distilled in four steps:\n\n- Extract claims, definitions, analytical tactics, examples, and source references while preserving their source identity.\n- Inventory the recurring decision questions before applying a product taxonomy.\n- Reconcile overlapping questions by analytical consequence, splitting items when their evidence needs or failure conditions differ.\n- Project the result into 65 consistent cards, eight modules, and three task workflows.\n\n```\nflowchart LR\n    corpus[\"<b>Source corpus</b><br/>primary writings, letters,<br/>interviews, filings\"]\n    extract[\"<b>1 · Extract</b><br/>source-native insights,<br/>identity preserved\"]\n    inventory[\"<b>2 · Inventory</b><br/>recurring decision questions,<br/>before any taxonomy\"]\n    reconcile[\"<b>3 · Reconcile</b><br/>overlapping questions merged<br/>or split by consequence\"]\n    project[\"<b>4 · Project</b><br/>consistent cards, modules,<br/>and task workflows\"]\n    corroborate[\"<b>Corroborate</b><br/>every card re-checked against<br/>the full landed collection\"]\n    validate[\"<b>Validate</b><br/>one command: contracts,<br/>sources, routing, links\"]\n    use[\"<b>Agent use</b><br/>router loads a small card<br/>set per task, never all\"]\n\n    corpus --> extract --> inventory --> reconcile --> project\n    project --> corroborate --> validate --> use\n\n    classDef src fill:#f5f0e1,stroke:#b8a878,color:#1f2328;\n    classDef step fill:#e9f2fc,stroke:#7ba7d7,color:#1f2328;\n    classDef check fill:#e3f3ee,stroke:#6fb39c,color:#1f2328;\n    classDef out fill:#f1ecfa,stroke:#a68fd8,color:#1f2328;\n    class corpus src\n    class extract,inventory,reconcile,project step\n    class corroborate,validate check\n    class use out\n```\n\nThe published cards provide representative, not exhaustive, coverage. The process prioritized recurring decision questions rather than attempting to map every passage in the source material.\n\nEach published card names one to three representative Buffett or Berkshire sources and the role each played: defining, supporting, implementing, illustrating, or qualifying the guidance, or specializing it for one context. These source notes explain the synthesis; they are not quotations, exhaustive literature reviews, independent corroboration, or a substitute for checking the original writing and the company evidence under analysis.\n\nAfter synthesis, every card was reconciled against the project’s full landed source\ncollection: Berkshire letters from 1977 through 2024, the 2025 transition and farewell\nletters, the partnership letters, the Owner’s Manual, the 50th-anniversary essays,\nBuffett’s later public letters and comments, *The Essays of Warren Buffett* arrangement,\nand the Lowenstein biography.\nEach card carries abbreviated corroboration citations naming additional checked\nlocations where its point is stated, applied, or qualified; the\n[source key](/finterm-ai/investment-skills/blob/main/skills/buffett-investment-framework/references/00-source-key.md) resolves\nevery abbreviation, explains the evidence character of each source, and gives a\nverified retrieval location for every citation key.\n\nThis project is not affiliated with, approved by, or endorsed by Warren Buffett or Berkshire Hathaway.\n\n**Deterministic routing:**`scripts/framework.py`\n\nis a standard-library Python script; no model chooses the card load. The same intent and topics always produce the same cards.**A hard card cap:** 5–12 cards per pass; agent context is treated as a budget.**A one-command validator:**`validate`\n\nchecks the 65-card namespace and per-card field contracts.**A structural authority boundary:** cards return supported, challenged, or unresolved findings.\n\nThe `npx skills add`\n\ncommand in [Try It](#try-it) installs into whichever agent\ndirectories the CLI detects.\nAdd `-g`\n\nto install for your user instead of the current project, `--copy`\n\nto install an\nindependent snapshot rather than a symlink, and `--list`\n\nto see what a repository offers\nwithout installing it.\nThe CLI attempts every agent it detects; a per-agent failure line for an agent you do\nnot use (for example, PromptScript declining global installs) leaves the successful\ninstalls intact.\n\nTo install without the CLI, copy the skill folder into the directory your agent reads. From this repository’s root:\n\n```\n# Cross-agent project install (Codex, Cursor, and pi read this path natively)\nmkdir -p your-project/.agents/skills\ncp -R skills/buffett-investment-framework your-project/.agents/skills/\n\n# Claude Code\nmkdir -p your-project/.claude/skills\ncp -R skills/buffett-investment-framework your-project/.claude/skills/\n\n# Codex\ncp -R skills/buffett-investment-framework \"${CODEX_HOME:-$HOME/.codex}/skills/\"\n```\n\nDevelopment docs live in [docs/development.md](/finterm-ai/investment-skills/blob/main/docs/development.md) and\n[viz/README.md](/finterm-ai/investment-skills/blob/main/viz/README.md).\n\nMIT. See [LICENSE](/finterm-ai/investment-skills/blob/main/LICENSE).\n\nThis project is open-source reference material, **not investment advice**. finterm.ai\nmakes no recommendation to buy, sell, or hold any security and accepts no liability for\ninvestment decisions you or your agents make.", "url": "https://wpnews.pro/news/investment-skills-lexicon-how-great-investors-think", "canonical_source": "https://github.com/finterm-ai/investment-skills", "published_at": "2026-07-22 17:04:44+00:00", "updated_at": "2026-07-22 17:22:55.304647+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-tools", "ai-agents", "large-language-models", "ai-products"], "entities": ["Finterm AI", "Warren Buffett", "Berkshire Hathaway", "Claude Code", "Codex", "Agent Skills", "Greg Abel", "Claude Fable"], "alternates": {"html": "https://wpnews.pro/news/investment-skills-lexicon-how-great-investors-think", "markdown": "https://wpnews.pro/news/investment-skills-lexicon-how-great-investors-think.md", "text": "https://wpnews.pro/news/investment-skills-lexicon-how-great-investors-think.txt", "jsonld": "https://wpnews.pro/news/investment-skills-lexicon-how-great-investors-think.jsonld"}}