The open OS for building an AI-native company in hard-mode sectors. A 15-chapter Handbook, a 90-term Dictionary, and 24 runnable skills for Claude Code, Codex, Cursor, and Gemini. CC-BY-4.0 / Apache-2.0.
Use when a founder says "how do I actually build this", "Claude Code/Codex/Cursor keeps making a mess", "the agent wrote spaghetti", "it passed but I don't trust it", "I'm just vibe-coding", "the agent went off and rewrote everything", "do I need tests for this", or is starting a feature and wants the agent on a…
Use when a founder is stuck in pilot purgatory, when deals go quiet after a thrilled champion says yes, when they say "the pilot stalled", "they loved the demo but procurement went dark", "how do I get from pilot to paid", "we keep automating spam", "scale sales without hiring", or they are about to put a pilot logo…
Use when a founder faces a hard, consequential call and is about to guess - build-or-buy, what to ship, whether to automate something irreversible, whether a slick demo proves anything - or says "should I build this", "is this defensible", "the demo works, can we ship", "talk me through this decision", "what would you…
Use when a founder says "let's start building", "what's the architecture", "should I just prototype this", "the codebase is a mess", "the AI keeps building a different thing each time", "it worked at ten files and broke at four hundred", "no one holds the system in their head", or an agent-generated prototype is…
Use when a real outcome just landed and the lesson should be banked so the OS stops repeating mistakes - an eval result, a customer reply, a metric move, a failed or won launch, a Share-of-Model move - when the founder says "we just learned", "that worked", "that failed", "log this", "note this for next time"…
Use when an AI system already exists or is half-designed and you need to judge whether it is AI-native or a dressed-up wrapper. Reach for it when a founder says "is this actually AI-native", "audit my architecture", "why does it feel fragile", "an investor will run the Remove-the-AI test on us", "the agent does…
Use when an AI or agent workload is heading into a security review, an enterprise buyer's questionnaire, or an audit and you do not yet know which frameworks you touch. Reach for this when you say "the customer's security team is asking", "do we need SOC 2", "is this HIPAA", "GDPR applies to us right?", "the EU AI…
Use when a founder is running customer interviews and wants the truth, not applause - when they say "everyone loves it", "I talked to fifteen people and they're all interested", "how do I validate this with users", "I fed my call notes to the model and it says strong demand", "they said they'd definitely use it", or…
Use when a founder wants to automate a recurring agentic task and is moving from prompting to system design - when they say "should I set up a loop", "automate this with agents", "run this nightly", "make the agent prompt itself", "my agent loops and burns tokens", "set up CI triage", or "automate the dependency…
Use when a founder is about to build the first version and is piling features instead of proving one loop - "what's my MVP", "what should I build first", "scope this down", "let's add X too", "the demo works, ship it", "which feature first", "this is our biggest use case so start there", or they are scoping the…
Use when a founder says "is this safe to ship", "it hallucinated to a customer", "it made up a citation", "we need clinical/food-safety/financial guardrails", "how do I evaluate this", "the demo worked but I'm scared to launch", "what if it's wrong about a dose", or is in health, food, or finance and an answer could…
Use when a founder has a vague AI idea and needs to sharpen it into one falsifiable bet before building - when they say "is this worth building", "how do I validate this", "I have an idea for an AI product", "the demo works, what now", "everyone I talk to says they love it", "how do I know if this is real", or are…
Use when one agent is trying to do every operational job and it is drifting, getting confused, or running up a bill - when the founder says "my agent does everything", "it routed a refund through the wrong logic", "it got stuck in a loop all weekend", "the prompt is 4000 words and nobody can touch it", "I got a…
Use when a founder asks why the AI engines never name them, when traffic holds but the buyer "already decided" before the call, when they say "we rank but ChatGPT cites a competitor", "how do I show up in Perplexity", "write the pillar page", "our content reads like everyone else's", or they are about to publish a…
Use when a founder has a pile of raw input - interview notes, support logs, search data, papers, competitor reviews - and needs to convert it into testable bets, or has too many ideas and cannot pick what to test first. Triggers: "I have all this research, now what", "what should I test first", "I've got ten…
Use when a founder needs the market, competitor, and regulatory terrain mapped before building - when they say "who are my competitors", "what's the regulatory path", "how long is EFSA or Novel Foods or CE", "is the market real", "what's the TAM", "where's the wedge", "we'll just be better than them", or are entering…
Use when a founder is mistaking AI novelty for demand and needs the real PMF read - "do we have PMF", "is this working", "what metrics matter", "users love the demo but", "we got 10,000 signups", "the launch went viral", "should we raise on this traction", or they are counting signups and total traffic instead of…
Use when a founder needs to know what actually defends the company once building is free - when they say "what's my moat", "a competitor could clone this in a weekend", "is my data a moat", "we collect a lot of data", "anyone can rent the same model", "we don't have enough data to compete", "is this defensible", or…
Use when an agent-built codebase has grown faster than anyone understands it and every change feels risky - when they say "the codebase is a mess", "I'm scared to touch this", "one fix broke three other things", "it worked when we built it", "we have a God Agent", "there's a ghost in the system", "tech debt is killing…
Use when an agent is about to meet real users or an auditor and you need to attack it first, the way a hostile user or a security reviewer would. Reach for this when you say "try to break it", "can someone jailbreak this", "is it injection-safe", "what if a user lies to it", "the agent has tools and I am scared"…
Use when an agent is about to get tools or connectors and you need to bound what it can actually do, the way a security reviewer would before granting access. Reach for this when you say "should I connect this MCP server", "what can my agent reach", "is this tool safe to expose", "it has access to our data now", "tool…
Use when an AI workflow has to survive a model that fails, stalls, or answers with low confidence instead of shipping a confident wrong answer - when they say "what happens when the model is down", "it gave a fluent wrong answer", "the API timed out and the whole thing broke", "a pilot user hit a dead end and stopped…
Use when a founder needs to know how often the AI engines name them versus rivals for the questions buyers actually ask, when they say "measure my Share of Model", "do ChatGPT and Perplexity cite us", "are we losing to a competitor in the answer box", "did the GEO work move the number", "set my baseline", or "track…
Use when a founder is unsure where to begin or what to do next on an AI build - when they say "where do I start", "I want to build with AI", "is my idea AI-native or a wrapper", "what's the next step", "should I add a billing page now", "I have an idea but no plan", "I'm overwhelmed", or are about to pour months into…