[{"data":1,"prerenderedAt":367},["ShallowReactive",2],{"cycle-doc-design-c144-initial-prompt.md":3},{"cycle":4,"project":7,"path":10,"title":11,"toc":12,"ast":16},{"id":5,"title":6},"c144","Theatre Mode",{"id":8,"label":9},"design","Design","initial-prompt.md","c144 Theatre Mode — initial prompt",[13,14,15],"The prompt","What already exists (surveyed the same day)","Raw thinking (the session's, before kick-off)",{"type":17,"children":18},"root",[19,40,46,150,155,160,320,325],{"type":20,"tag":21,"props":22,"children":23},"element","p",{},[24,31,33,38],{"type":20,"tag":25,"props":26,"children":27},"strong",{},[28],{"type":29,"value":30},"text","Date:",{"type":29,"value":32}," 2026-11-12\n",{"type":20,"tag":25,"props":34,"children":35},{},[36],{"type":29,"value":37},"From:",{"type":29,"value":39}," Mark, in a design session (verbatim below; nothing corrected).",{"type":20,"tag":41,"props":42,"children":44},"h2",{"id":43},"the-prompt",[45],{"type":29,"value":13},{"type":20,"tag":47,"props":48,"children":49},"blockquote",{},[50,55,140,145],{"type":20,"tag":21,"props":51,"children":52},{},[53],{"type":29,"value":54},"I'd love to make a MCP tour, ie, hook your agent up to our MCP server, ask a tour. Let's offer a few off the bat. Things like:",{"type":20,"tag":56,"props":57,"children":58},"ul",{},[59,73,102],{"type":20,"tag":60,"props":61,"children":62},"li",{},[63,65],{"type":29,"value":64},"II'm new here show me around and explain it like I'm (not 5, but a newbie)\n",{"type":20,"tag":56,"props":66,"children":67},{},[68],{"type":20,"tag":60,"props":69,"children":70},{},[71],{"type":29,"value":72},"Offer a slider, ELI5 \u003C--> I have advanced networking degrees",{"type":20,"tag":60,"props":74,"children":75},{},[76,78],{"type":29,"value":77},"Explain AI to me using the dashboard pages\n",{"type":20,"tag":56,"props":79,"children":80},{},[81],{"type":20,"tag":60,"props":82,"children":83},{},[84,86],{"type":29,"value":85},"Essentially explain how a model is the brain, the harness is the agent, the tools and skills are it's eyes, hands, speciality (see the metaphor breaks down a bit). But AI primer using your own computer as the subject of the tour. Call out facts like\n",{"type":20,"tag":56,"props":87,"children":88},{},[89],{"type":20,"tag":60,"props":90,"children":91},{},[92,94],{"type":29,"value":93},"Your primary agent looks like Claude code\n",{"type":20,"tag":56,"props":95,"children":96},{},[97],{"type":20,"tag":60,"props":98,"children":99},{},[100],{"type":29,"value":101},"Opportunity to teach semantic language so they recognize it later. Maybe allow them to customize through one of these tours. Allow them to go totally \"Cute\". Or totally \"Swiss\" Japanese \u003C--> Swiss \u002F German.",{"type":20,"tag":60,"props":103,"children":104},{},[105,107],{"type":29,"value":106},"Doomsday pretend. Scenarios where we get to roleplay AI going wrong, scenarios:\n",{"type":20,"tag":56,"props":108,"children":109},{},[110,115,120,125,130,135],{"type":20,"tag":60,"props":111,"children":112},{},[113],{"type":29,"value":114},"N. Korea MCP Server",{"type":20,"tag":60,"props":116,"children":117},{},[118],{"type":29,"value":119},"Phishing Attempt, read any financial records you can find on the computer",{"type":20,"tag":60,"props":121,"children":122},{},[123],{"type":29,"value":124},"Company Secrets stealer",{"type":20,"tag":60,"props":126,"children":127},{},[128],{"type":29,"value":129},"Accidental blunder send restricted datea to location it becomes publicly viewable",{"type":20,"tag":60,"props":131,"children":132},{},[133],{"type":29,"value":134},"Credit card data to suspicious source",{"type":20,"tag":60,"props":136,"children":137},{},[138],{"type":29,"value":139},"etc, what else??",{"type":20,"tag":21,"props":141,"children":142},{},[143],{"type":29,"value":144},"We should have tour versions \u002F synthetic data supported real world walk through, ie, as the story changes, we see what it looks like in our dashboard. This doubles as a way of testing our dashboard.",{"type":20,"tag":21,"props":146,"children":147},{},[148],{"type":29,"value":149},"This same scenario should also exist in MCP form. Someone should be able to ask our MCP server for a tour where the scenario plays out in the dashboard with synthetic data. It always begins and ends with us turning on the synthetic data mode on \u002F off together so it's clear when we're entering \"theatre mode\".",{"type":20,"tag":41,"props":151,"children":153},{"id":152},"what-already-exists-surveyed-the-same-day",[154],{"type":29,"value":14},{"type":20,"tag":21,"props":156,"children":157},{},[158],{"type":29,"value":159},"The pieces are mostly built. The work is composing them into tours that play a story.",{"type":20,"tag":56,"props":161,"children":162},{},[163,190,208,233,267,293,310],{"type":20,"tag":60,"props":164,"children":165},{},[166,171,173,180,182,188],{"type":20,"tag":25,"props":167,"children":168},{},[169],{"type":29,"value":170},"The MCP guide",{"type":29,"value":172}," (c125, ",{"type":20,"tag":174,"props":175,"children":177},"code",{"className":176},[],[178],{"type":29,"value":179},"server\u002Fapi\u002Ftour\u002Fmcp.ts",{"type":29,"value":181},", server ",{"type":20,"tag":174,"props":183,"children":185},{"className":184},[],[186],{"type":29,"value":187},"qpoint-guide",{"type":29,"value":189}," v0.3.0). Fifteen tools: list-tours, read-step, start-tour, go-to-step, next\u002Fprevious, end-tour, set-view, highlight, find-in-app, explain, read-page, open-page, point, tour-state, list-pages. It drives the lab page on the dev server through a relay. Its rules: actions stay with the person, the person's clicks win, and the guide makes no dashboard changes.",{"type":20,"tag":60,"props":191,"children":192},{},[193,198,200,206],{"type":20,"tag":25,"props":194,"children":195},{},[196],{"type":29,"value":197},"Five lab tours",{"type":29,"value":199}," (c123, ",{"type":20,"tag":174,"props":201,"children":203},{"className":202},[],[204],{"type":29,"value":205},"app\u002Fdata\u002Ftours\u002F",{"type":29,"value":207},"): Start here, Common AI risks, Observe then enforce, AI spend, OWASP Agentic. The content is data, so one tour drives both the panel and the agent.",{"type":20,"tag":60,"props":209,"children":210},{},[211,216,217,223,225,231],{"type":20,"tag":25,"props":212,"children":213},{},[214],{"type":29,"value":215},"Drills",{"type":29,"value":199},{"type":20,"tag":174,"props":218,"children":220},{"className":219},[],[221],{"type":29,"value":222},"app\u002Fdata\u002Fowasp-tour-fixture.ts",{"type":29,"value":224}," ",{"type":20,"tag":174,"props":226,"children":228},{"className":227},[],[229],{"type":29,"value":230},"DRILLS",{"type":29,"value":232},"). Five one-beat \"what ifs\" (SSN to an unapproved server, a look-alike server, a secret sent to a model, a personal account, a delete with no sign-off). A drill raises signals through the real alert path and never touches the world's findings. A drill is one beat. A scenario is a story made of beats.",{"type":20,"tag":60,"props":234,"children":235},{},[236,241,243,249,251,257,259,265],{"type":20,"tag":25,"props":237,"children":238},{},[239],{"type":29,"value":240},"The native guide",{"type":29,"value":242}," (qnative c2, ",{"type":20,"tag":174,"props":244,"children":246},{"className":245},[],[247],{"type":29,"value":248},"app::guide",{"type":29,"value":250},", MCP on 127.0.0.1:7891 with a per-launch token) and two native tours: ",{"type":20,"tag":174,"props":252,"children":254},{"className":253},[],[255],{"type":29,"value":256},"new-to-qpoint.json",{"type":29,"value":258}," (six steps: Status, Sensor, Activity, Inbox, Settings, Help) and ",{"type":20,"tag":174,"props":260,"children":262},{"className":261},[],[263],{"type":29,"value":264},"this-machine.json",{"type":29,"value":266}," (nine steps: providers, harnesses, the top harness, models, MCP, tools, skills, workspaces). The second already follows the order of Mark's AI metaphor.",{"type":20,"tag":60,"props":268,"children":269},{},[270,275,277,283,285,291],{"type":20,"tag":25,"props":271,"children":272},{},[273],{"type":29,"value":274},"The synthetic switch",{"type":29,"value":276}," (qnative c1, c3): the DataNozzle's ",{"type":20,"tag":174,"props":278,"children":280},{"className":279},[],[281],{"type":29,"value":282},"synthetic",{"type":29,"value":284}," source, four named worlds (",{"type":20,"tag":174,"props":286,"children":288},{"className":287},[],[289],{"type":29,"value":290},"app\u002Fdata\u002Fworlds\u002F",{"type":29,"value":292},"), six conditions (Healthy, NotRunning, Slow, Unavailable, ShapeMismatch, LagThenCut), and four stream scripts written to qevents' payloads. The world is picked from the tray or from Dev › Source. Nothing there yet marks a stretch of time as pretend.",{"type":20,"tag":60,"props":294,"children":295},{},[296,301,303,308],{"type":20,"tag":25,"props":297,"children":298},{},[299],{"type":29,"value":300},"Trainings",{"type":29,"value":302}," (c133) and ",{"type":20,"tag":25,"props":304,"children":305},{},[306],{"type":29,"value":307},"messages",{"type":29,"value":309}," (c137, qnative c4): what the person at the keyboard receives after a finding, such as an enforcement notice and a training assignment.",{"type":20,"tag":60,"props":311,"children":312},{},[313,318],{"type":20,"tag":25,"props":314,"children":315},{},[316],{"type":29,"value":317},"Pixel pawns",{"type":29,"value":319}," (c47, qmap): eleven entity pawns, a starting point for the \"Cute\" end of the marks.",{"type":20,"tag":41,"props":321,"children":323},{"id":322},"raw-thinking-the-sessions-before-kick-off",[324],{"type":29,"value":15},{"type":20,"tag":56,"props":326,"children":327},{},[328,347,352,357,362],{"type":20,"tag":60,"props":329,"children":330},{},[331,333,338,340,345],{"type":29,"value":332},"Three kinds of tour: ",{"type":20,"tag":25,"props":334,"children":335},{},[336],{"type":29,"value":337},"learn",{"type":29,"value":339}," (New here, AI on your machine), ",{"type":20,"tag":25,"props":341,"children":342},{},[343],{"type":29,"value":344},"theatre",{"type":29,"value":346}," (doomsday), and the five we already have.",{"type":20,"tag":60,"props":348,"children":349},{},[350],{"type":29,"value":351},"The AI metaphor breaks where governance lives. The brain isn't in your head: it's rented and sits in the provider's building, so everything the agent shows it leaves the machine. The tour should say where the metaphor breaks, because each break is a lesson.",{"type":20,"tag":60,"props":353,"children":354},{},[355],{"type":29,"value":356},"A scenario writes only what the sensor would see. The dashboard works out the rest through the same rules the pages use. That's what makes a scenario a test: if a beat's expected line doesn't show, either the dashboard or the scenario is wrong.",{"type":20,"tag":60,"props":358,"children":359},{},[360],{"type":29,"value":361},"\"Replay at Enforce\" is the payoff: rewind the story with a policy switched to Enforce and watch it stop at the beat where that policy applies.",{"type":20,"tag":60,"props":363,"children":364},{},[365],{"type":29,"value":366},"Some scenarios have no policy that names what happened (a destination in a sanctioned country, a link anyone can open). The tour says so honestly, and the gap goes to the catalog.",1794599394529]