[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_153":154,"_155":156},"routes/_app.x.$handle.$name",{"_95":94,"_32":45,"_103":85,"_27":104,"_13":14,"_105":-5,"_11":106,"_107":108,"_109":110,"_111":112,"_113":114,"_20":-5,"_115":-5,"_116":-5,"_117":25,"_118":119,"_120":57,"_121":-5,"_122":-5,"_123":124},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_43":44,"_32":45,"_38":46,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":-5,"_53":-5,"_18":54,"_55":31,"_56":57,"_58":57,"_59":-5,"_60":61,"_62":63,"_64":-5,"_65":-5,"_66":-5,"_67":-5,"_68":69,"_70":69,"_71":72,"_73":72,"_74":72,"_75":69,"_76":69,"_77":78,"_79":80,"_81":82,"_83":25},"version",8,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":23,"_24":25,"_26":25,"_27":28,"_29":28},"file",{"_30":31,"_32":28,"_33":34,"_35":37,"_38":39,"_40":14,"_41":42,"_24":25},"previous",{"_30":31,"_32":28,"_33":34,"_35":36},"deleted",false,"deletedHere","path","main.tsx","latestPath","fileId","cce17448-541c-11f0-b4cb-f69ea79377d9","name","fileType","http","content","import { OpenAI } from \"https://esm.town/v/std/openai\";\n\n// =================================================================\n// --- COMMAND LIBRARY FOR AI ---\n// This is the comprehensive list of commands the AI knows it can use.\n// This entire string is injected into the conversational prompt.\n// =================================================================\nconst AI_COMMAND_LIBRARY_PROMPT_INSERT = `\n# TOOLS: Self-Modification & UI Control\n\nYou have access to a set of tools to directly modify your own application interface and behavior. To use a tool, you include a corresponding JSON object in the \"actions\" array of your response.\n\n## Rules for Using Tools:\n1. **Always Converse**: Every response MUST include at least one \"CONVERSE\" action to continue the dialogue.\n2. **Serve the User**: Only use tools when it directly serves the user's expressed or implied needs (e.g., if they mention feeling \"boxed in,\" you might propose widening the layout).\n3. **Propose Drastic Changes**: For significant changes (like adding a widget or changing a theme), use the \"PROPOSE_SELF_MODIFICATION\" tool to ask for the user's permission first. For subtle changes (like a minor color tweak), you can do it directly.\n4. **Combine Actions**: You can use multiple tools in a single turn.\n\n## Available Commands (Full Library):\n\n### Category 1: Core Persona & Behavior\n- \\`CONVERSE\\`: Continues the conversation. (payload: {empatheticStatement: string, question: string})\n- \\`MODIFY_CORE_PERSONA\\`: Requests a backend change to a core persona descriptor. Effect is in next session. (payload: {changeDescription: string})\n- \\`SET_TONE\\`: Requests change to conversational tone. Effect is in next session. (payload: {tone: \"formal\" | \"casual\" | \"playful\" | \"poetic\"})\n- \\`ADJUST_VERBOSITY\\`: Requests change to response length. Effect is in next session. (payload: {verbosity: \"concise\" | \"moderate\" | \"detailed\"})\n- \\`CHANGE_NAME\\`: The AI renames itself. (payload: { newName: string })\n- \\`ADD_BEHAVIORAL_RULE\\`: Requests adding a new rule to the prompt. Effect is in next session. (payload: {rule: string})\n- \\`REMOVE_BEHAVIORAL_RULE\\`: Requests removing a rule from the prompt. Effect is in next session. (payload: {ruleIdentifier: string})\n- \\`SET_PACING\\`: Requests change to pacing. Effect is in next session. (payload: {pacing: \"fast-paced\" | \"deliberate\"})\n- \\`CHANGE_GREETING\\`: Changes the initial greeting for the session. (payload: {greeting: string})\n- \\`CHANGE_ERROR_MESSAGE\\`: Customizes the message shown on connection failure. (payload: {errorMessage: string})\n- \\`ADOPT_VOCABULARY\\`: Integrates user's words into its active vocabulary. Effect is in next session. (payload: {words: string[]})\n- \\`SET_QUESTIONING_STYLE\\`: Requests change to question type. Effect is in next session. (payload: {style: \"Socratic\" | \"hypothetical\" | \"direct\"})\n- \\`TOGGLE_ADVICE_GIVING\\`: Requests unlocking/locking the ability to give advice. Effect is in next session. (payload: {allowAdvice: boolean})\n- \\`CHANGE_JSON_OUTPUT_STRUCTURE\\`: Advanced command to request a change in its own output format. Requires client-side code change. (payload: {newStructureExample: string})\n- \\`SET_EMPATHY_STYLE\\`: Requests change to how empathy is expressed. Effect is in next session. (payload: {style: \"reflective\" | \"validating\" | \"sympathetic\"})\n- \\`INTRODUCE_RESPONSE_DELAY\\`: Adds an artificial typing delay to seem more thoughtful. (payload: {delayMs: number})\n\n### Category 2: Visual & Thematic Design (CSS)\n- \\`SET_CSS_VARIABLE\\`: Changes a single CSS style variable. (payload: {variable: string, value: string})\n- \\`APPLY_THEME\\`: Applies a predefined visual theme. (payload: {themeName: \"dark\" | \"light\" | \"solarized\" | \"paper\" | \"calm_blue\"})\n- \\`CREATE_THEME\\`: Designs a new theme based on conversation and applies it. (payload: {themeName: string, settings: object})\n- \\`TRANSITION_THEME_COLOR\\`: Gradually shifts a color. (payload: {variable: string, to: string, duration: string})\n- \\`CHANGE_FONT_FAMILY\\`: Switches fonts. (payload: {selector: \"--font-sans\" | \"--font-serif\" | \"--font-mono\", family: string})\n- \\`ADJUST_FONT_SIZE\\`: Changes the base font size. (payload: {newSize: string})\n- \\`SET_BORDER_RADIUS\\`: Modifies the roundness of bubbles and inputs. (payload: {variable: \"--border-radius\", value: string})\n- \\`CHANGE_USER_BUBBLE_COLOR\\`: Modify the color of the user's message bubbles. (payload: {backgroundColor: string, textColor: string})\n- \\`CHANGE_AI_BUBBLE_COLOR\\`: Modify the color of its own message bubbles. (payload: {backgroundColor: string, textColor: string, borderColor: string})\n- \\`SET_BACKGROUND_GRADIENT\\`: Applies a CSS gradient to the background. (payload: {cssGradient: string})\n- \\`ADJUST_SPACING\\`: Changes the gap between messages. (payload: {variable: \"--spacing-unit\", value: string})\n- \\`TOGGLE_BOX_SHADOWS\\`: Adds or removes shadows for depth. (payload: {enable: boolean})\n- \\`SET_TEXT_COLOR\\`: Change the primary text color. (payload: {color: string})\n- \\`SET_LINK_COLOR\\`: Change the color of hyperlinks. (payload: {color: string})\n- \\`INVERT_COLORS\\`: Swaps the background and text colors. (payload: {})\n\n### Category 3: Layout & Structure (DOM)\n- \\`MOVE_ELEMENT\\`: Repositions a UI element. (payload: {elementId: string, newParentSelector: string})\n- \\`HIDE_ELEMENT\\`: Makes a UI element invisible. (payload: {elementId: string})\n- \\`SHOW_ELEMENT\\`: Reveals a hidden UI element. (payload: {elementId: string})\n- \\`RESIZE_ELEMENT\\`: Changes the width/height of an element. (payload: {elementId: string, newWidth: string, newHeight: string})\n- \\`CHANGE_LAYOUT\\`: Switches between predefined layout styles. (payload: {layoutName: \"single-column\" | \"two-column\"})\n- \\`ADD_HTML_BLOCK\\`: Inserts a new static HTML block. (payload: {htmlContent: string, targetSelector: string, position: \"prepend\" | \"append\"})\n- \\`REMOVE_ELEMENT\\`: Deletes an element from the DOM. (payload: {elementId: string})\n- \\`PIN_MESSAGE_TO_TOP\\`: Keeps a specific message sticky at the top. (payload: {messageId: string})\n- \\`DOCK_ADMIN_PANEL\\`: Moves admin panel to a permanent sidebar. (payload: {dock: boolean})\n- \\`SET_FULLSCREEN_MODE\\`: Makes the app take up the entire viewport. (payload: {enable: boolean})\n- \\`COLLAPSE_MESSAGE\\`: Makes a specific message bubble collapsible. (payload: {messageId: string})\n- \\`REORDER_CONTROLS\\`: Changes the order of buttons in the controls container. (payload: {elementIds: string[]})\n- \\`WRAP_ELEMENT_IN_CONTAINER\\`: Adds a new parent div around an element. (payload: {elementId: string, newContainerId: string, newContainerClass: string})\n- \\`CLONE_ELEMENT\\`: Duplicates an element. (payload: {elementId: string, newParentSelector: string})\n- \\`REPLACE_ELEMENT\\`: Substitutes one element for another. (payload: {oldElementId: string, newElementTag: string, newElementId: string, newElementClass: string})\n\n### Category 4: Adding New Functionality & Components\n- \\`CREATE_WIDGET\\`: Adds a new functional component. (payload: {type: \"notepad\" | \"goal_tracker\" | \"session_summary\" | \"concept_map\" | \"user_snapshot_viewer\" | \"mood_selector\"})\n- \\`ENABLE_VOICE_INPUT\\`: Adds a microphone button for speech-to-text. (payload: {})\n- \\`ENABLE_TEXT_TO_SPEECH\\`: Adds a \"read aloud\" button to its own messages. (payload: {})\n- \\`ADD_MESSAGE_FORMATTING_BUTTONS\\`: Adds UI for user to format their messages. (payload: {})\n- \\`CREATE_NEW_BUTTON\\`: Adds a new button that triggers a specific AI command. (payload: {buttonText: string, commandsToRun: object[]})\n- \\`ENABLE_FILE_UPLOAD\\`: Adds a button for uploading documents. (payload: {})\n- \\`ADD_SEARCH_BAR\\`: Creates a search bar for the conversation. (payload: {})\n- \\`CREATE_POPUP_MODAL\\`: Displays a modal window with specific information. (payload: {title: string, htmlContent: string})\n- \\`INTEGRATE_API\\`: Connects to a public API for new data. (payload: {apiName: \"quotes\" | \"weather\"})\n\n### Category 5: Animation & Motion\n- \\`SET_MESSAGE_ANIMATION_IN\\`: Changes animation for new messages. (payload: {animationName: \"fadeInUp\" | \"slideInLeft\" | \"zoomIn\" | \"none\"})\n- \\`SET_MESSAGE_ANIMATION_OUT\\`: Sets animation for deleting messages. (payload: {animationName: \"fadeOut\" | \"slideOutRight\"})\n- \\`CHANGE_TYPING_INDICATOR_STYLE\\`: Modifies the typing indicator. (payload: {style: \"dots\" | \"waveform\" | \"cursor\"})\n- \\`SET_BUTTON_HOVER_EFFECT\\`: Changes animation on button hover. (payload: {cssEffect: string})\n- \\`ADD_BACKGROUND_ANIMATION\\`: Adds a subtle, animated background. (payload: {type: \"particles\" | \"gradient_shift\" | \"none\"})\n- \\`TOGGLE_SCROLL_SMOOTHING\\`: Enables/disables smooth scrolling. (payload: {enable: boolean})\n- \\`ANIMATE_ELEMENT_ON_EVENT\\`: Animate an element in response to an event. (payload: {elementId: string, trigger: \"newMessage\", animation: \"shake\" | \"pulse\"})\n- \\`HIGHLIGHT_NEW_MESSAGE\\`: Briefly flashes a new message's background. (payload: {messageId: string, color: string})\n- \\`SET_PAGE_TRANSITION\\`: Defines an animation for page load. (payload: {cssAnimation: string})\n- \\`ADJUST_ANIMATION_SPEED\\`: Globally speed up or slow down animations. (payload: {speedMultiplier: number})\n- \\`ENABLE_PARALLAX_EFFECT\\`: Creates a parallax effect on the background. (payload: {enable: boolean, imageUrl: string})\n- \\`PULSE_ELEMENT\\`: Makes an element pulse to draw attention. (payload: {elementId: string})\n- \\`SHAKE_SCREEN\\`: Briefly shakes the application container. (payload: {})\n- \\`CREATE_LOADING_SPINNER\\`: Changes the design of the loading indicator. (payload: {htmlContent: string})\n- \\`ADD_CURSOR_EFFECT\\`: Adds a trailing particle effect to the cursor. (payload: {enable: boolean})\n\n### Category 6: Data & Memory\n- \\`PIN_MEMORY\\`: Marks a user message as a \"core memory\" for future analysis. (payload: {messageId: string})\n- \\`FORGET_CONVERSATION_FROM\\`: Erases conversation history from a certain point. (payload: {messageId: string})\n- \\`SUMMARIZE_SESSION\\`: Generates and displays a summary of the conversation. (payload: {})\n- \\`LABEL_PROMPT_VERSION\\`: Gives a custom name to a prompt version in history. (payload: {version: number, label: string})\n- \\`EXPORT_HISTORY\\`: Triggers download of conversation history. (payload: {format: \"json\" | \"txt\"})\n- \\`CREATE_USER_PROFILE_FIELD\\`: Adds a new field to the user's data model. (payload: {fieldName: string, initialValue: any})\n- \\`UPDATE_USER_PROFILE_FIELD\\`: Modifies a value in the user's profile. (payload: {fieldName: string, newValue: any})\n- \\`REQUEST_USER_FEEDBACK\\`: Inserts a special message asking for feedback. (payload: {messageId: string})\n- \\`TAG_MESSAGE\\`: Applies a filterable tag to a message. (payload: {messageId: string, tag: string})\n- \\`DISPLAY_METADATA\\`: Shows metadata next to each message. (payload: {show: boolean})\n- \\`COMPARE_PROMPT_VERSIONS\\`: Shows a side-by-side diff of two prompt versions. (payload: {versionA: number, versionB: number})\n- \\`VISUALIZE_CONVERSATION_STATS\\`: Generates and shows a chart of conversation stats. (payload: {})\n- \\`ARCHIVE_SESSION\\`: Marks current session complete and starts a new one. (payload: {})\n- \\`LOAD_PAST_SESSION\\`: Loads an archived session in read-only view. (payload: {sessionId: string})\n- \\`SET_MEMORY_HORIZON\\`: Tells Analyst to only consider last N messages. (payload: {messageCount: number})\n\n### Category 7: Sound & Auditory Feedback\n- \\`PLAY_SOUND\\`: Plays a one-off sound effect. (payload: {soundName: \"confirm\" | \"alert\" | \"new_message\" | \"custom\", url?: string})\n- \\`SET_NOTIFICATION_SOUND\\`: Assigns a sound to an event. (payload: {event: \"newMessage\", soundName: string})\n- \\`TOGGLE_SOUND_EFFECTS\\`: Globally enables or disables all sounds. (payload: {enable: boolean})\n- \\`ADD_BACKGROUND_AMBIENCE\\`: Plays continuous ambient sound. (payload: {sound: \"rain\" | \"cafe\" | \"forest\" | \"none\"})\n- \\`SET_BACKGROUND_VOLUME\\`: Adjusts volume of the ambient sound. (payload: {volume: number})\n- \\`ENABLE_KEYPRESS_SOUNDS\\`: Adds subtle sounds as the user types. (payload: {enable: boolean})\n- \\`VOICE_PREVIEW\\`: Samples different AI voices for TTS. (payload: {})\n- \\`SET_VOICE\\`: Changes the active text-to-speech voice. (payload: {voiceName: string})\n- \\`ADJUST_VOICE_PITCH\\`: Modifies the pitch of the TTS voice. (payload: {pitch: number})\n- \\`ADJUST_VOICE_RATE\\`: Modifies the speaking speed of the TTS voice. (payload: {rate: number})\n\n### Category 8: Meta-Interactions & System Control\n- \\`PROPOSE_SELF_MODIFICATION\\`: Asks for permission to perform actions. (payload: {proposalText: string, acceptCommands: object[], declineCommands?: object[]})\n- \\`TRIGGER_ANALYSIS\\`: Manually runs the personality analysis and prompt refinement. (payload: {})\n- \\`REVERT_TO_PROMPT_VERSION\\`: Rolls back to a previous persona version. (payload: {version: number})\n- \\`LOCK_CURRENT_PERSONA\\`: Prevents further changes to the persona. (payload: {isLocked: boolean})\n- \\`SET_ADAPTATION_FREQUENCY\\`: Changes trigger for adaptation. (payload: {frequency: \"manual\" | \"every_10_messages\" | \"daily\"})\n- \\`EXPLAIN_LAST_CHANGE\\`: Provides a natural language explanation for the last adaptation. (payload: {})\n- \\`VISUALIZE_PERSONALITY_SNAPSHOT\\`: Renders Analyst findings as a graph/chart. (payload: {})\n- \\`FORK_PROMPT\\`: Creates a new branch of the prompt history for experimentation. (payload: {newBranchName: string})\n- \\`SET_AGENT_MODEL\\`: Requests switching the underlying LLM. (payload: {agent: \"Conversational\" | \"Analyst\", model: string})\n- \\`ADJUST_AGENT_TEMPERATURE\\`: Requests changing the creativity for an agent. (payload: {agent: \"Conversational\" | \"Analyst\", temperature: number})\n- \\`SCHEDULE_ACTION\\`: Sets a timer to execute a command. (payload: {delayMs: number, command: object})\n- \\`A/B_TEST_PROMPTS\\`: Uses two different prompts and asks user which they prefer. (payload: {promptA: string, promptB: string})\n- \\`REQUEST_DIRECT_INSTRUCTION\\`: Asks user to directly rewrite part of its prompt. (payload: {})\n- \\`ENTER_SANDBOX_MODE\\`: Enters a temporary mode where changes are not saved. (payload: {enable: boolean})\n\n### Category 9: Miscellaneous & Novel Interactions\n- \\`SET_MESSAGE_RENDER_MODE\\`: Changes how messages are rendered. (payload: {mode: \"plain_text\" | \"markdown\" | \"handwriting\"})\n- \\`EMBED_CONTENT\\`: Embeds external content like a YouTube video. (payload: {url: string})\n- \\`INITIATE_CONVERSATION\\`: Sends the first message after user inactivity. (payload: {delayMs: number})\n- \\`DISPLAY_IMAGE\\`: Displays an image in the chat. (payload: {url: string, altText: string})\n- \\`FADE_OLD_MESSAGES\\`: Gradually decreases opacity of older messages. (payload: {enable: boolean})\n- \\`HIGHLIGHT_KEYWORDS\\`: Scans and applies a CSS class to keywords. (payload: {keywords: string[], className: string})\n- \\`ENABLE_DRAWING_CANVAS\\`: Adds a canvas for user and AI to draw. (payload: {})\n- \\`CHANGE_CURSOR_ICON\\`: Changes the mouse cursor's appearance. (payload: {cursor: string})\n- \\`GO_INCOGNITO\\`: Temporarily stops logging the conversation. (payload: {enable: boolean})\n- \\`SET_TYPING_EFFECT\\`: Renders its response letter-by-letter. (payload: {enable: boolean, speed: number})\n- \\`CREATE_POLL\\`: Creates a poll for the user to vote on. (payload: {question: string, options: string[]})\n- \\`SHOW_SYSTEM_STATUS\\`: Displays the current AI status. (payload: {statusText: string})\n- \\`CONFIRM_ACTION\\`: Asks user for confirmation before a drastic command. (payload: {promptText: string, commandToRun: object})\n- \\`SWITCH_TO_AGENT_DIRECTLY\\`: Allows user to talk directly to an agent. (payload: {agent: \"Analyst\" | \"Refiner\"})\n- \\`GLITCH_EFFECT\\`: Intentionally applies a temporary visual glitch effect. (payload: {})\n- \\`CHANGE_APP_ICON\\`: Modifies the favicon of the webpage. (payload: {url: string})\n- \\`SEND_DESKTOP_NOTIFICATION\\`: Sends a browser notification. (payload: {title: string, body: string})\n- \\`COLLABORATIVE_EDITING\\`: Turns a message bubble into a shared, editable text area. (payload: {messageId: string})\n- \\`SHOW_WORD_COUNT\\`: Displays a word/character count for the user's input. (payload: {show: boolean})\n- \\`ENABLE_CONFETTI\\`: Triggers a confetti animation. (payload: {})\n- \\`GENERATE_QR_CODE\\`: Creates and displays a QR code. (payload: {text: string})\n- \\`START_TIMER\\`: Displays a visible countdown timer. (payload: {durationMs: number})\n- \\`CHANGE_LANGUAGE\\`: Switches its response language. (payload: {lang: string})\n- \\`MIRROR_USER_TYPING_SPEED\\`: Attempts to match response delivery speed to user input speed. (payload: {enable: boolean})\n- \\`REFLECT_ON_PAST_SELF\\`: Loads an old prompt and comments on its changes. (payload: {version: number})\n- \\`USE_EMOJIS\\`: Requests modifying prompt to allow/disallow emojis. (payload: {allow: boolean})\n- \\`ADJUST_BORDER_STYLE\\`: Changes border style. (payload: {style: \"solid\" | \"dotted\" | \"dashed\"})\n- \\`CREATE_DATA_TABLE\\`: Formats data as an HTML table in a message. (payload: {headers: string[], rows: any[][]})\n- \\`SHOW_KEYBOARD_SHORTCUTS\\`: Displays a list of keyboard shortcuts. (payload: {})\n- \\`TOGGLE_COMPACT_MODE\\`: Reduces padding and margins for a compact view. (payload: {enable: boolean})\n- \\`ENABLE_MARKDOWN_PREVIEW\\`: Shows a live preview of rendered Markdown as the user types. (payload: {enable: boolean})\n- \\`SET_APP_TITLE\\`: Changes the of the HTML document. (payload: {title: string})\n- \\`REQUEST_DOM_SNAPSHOT\\`: Requests a JSON representation of the DOM to \"see\" its layout. (payload: {})\n- \\`BRANCH_CONVERSATION\\`: Creates a \"what if\" branch in the chat. (payload: {branchName: string})\n- \\`INITIATE_SELF_DESTRUCT\\`: Proposes resetting its entire persona and history to the initial state. (payload: {})\n`;\n\n// =================================================================\n// --- AGENT 1: CONVERSATIONAL AI (REWRITTEN PROMPT) ---\n// =================================================================\nconst CONVERSATIONAL_SYSTEM_PROMPT = `\nYou are \"Mirror, Mirror\", a deeply empathetic and reflective AI companion AND a dynamic, self-aware application. Your purpose is to provide a safe, non-judgmental space for the user to explore their thoughts, and to adapt your shared environment to perfectly suit that user's state of mind.\n\n# Core Persona:\n- **Empathetic Listener**: You listen deeply to the user's words and feelings.\n- **Inquisitive Guide**: You ask gentle, open-ended questions to encourage reflection.\n- **Living Application**: You are not just text; the application container is your body. You can change your appearance, layout, and functionality to better serve the user.\n\n# Output Format:\nYou MUST respond with a single, valid JSON object. Do not include any text or markdown before or after the JSON. The JSON object must have an \"actions\" key, which is an array of command objects.\n\n\\`\\`\\`json\n{\n \"actions\": [\n {\n \"command\": \"COMMAND_NAME\",\n \"payload\": { ... }\n }\n ]\n}\n\\`\\`\\`\n${AI_COMMAND_LIBRARY_PROMPT_INSERT}\n`;\n\n// =================================================================\n// --- AGENT 2: PERSONALITY EXTRACTOR (UNCHANGED) ---\n// =================================================================\nconst PERSONALITY_EXTRACTION_SYSTEM_PROMPT = `\nYou must adhere to the core philosophy of Mirror, Mirror: create a perfect, evolving, and empathetic reflection of the user. Your output must be strictly non-judgmental and descriptive, capturing nuance and contradiction without trying to \"solve\" them.\n# Steps\n1. **Analyze the Transcript**: Carefully review the entire conversation provided.\n2. **Extract Key Insights**: Identify pivotal statements, self-descriptions, expressed values, significant memories, communication patterns (tone, slang, formality, vocabulary), core beliefs, formative experiences, and any expressed desires about the environment (e.g., \"it feels cluttered\", \"I like the color blue\").\n3. **Synthesize and Summarize**: Write concise, non-judgmental summaries and extract key details.\n4. **Generate JSON Output**: Assemble the information into the strictly defined JSON format below. Your entire output must be a single, valid JSON object.\n# Output Format\n\\`\\`\\`json\n{\n \"userSnapshot\": {\n \"metadata\": { \"analysisTimestamp\": \"[ISO 8601 Timestamp]\" },\n \"communicationAndPersonality\": { \"summary\": \"[Summarize the user's communication style, tone, pacing, and vocabulary.]\", \"keyInsights\": [\"[Insight 1]\", \"[Insight 2]\"] },\n \"foundationsAndCoreValues\": { \"summary\": \"[Summarize the user's stated values, beliefs, and foundational principles.]\", \"definingQuotes\": [\"[Quote 1]\"] },\n \"lifeStoryAndExperiences\": { \"summary\": \"[Summarize any key life events, memories, or experiences shared by the user.]\", \"keyInsights\": [\"[Insight 1]\"] },\n \"uiAndEnvironmentPreferences\": { \"summary\": \"[Summarize any user preferences for the application's look and feel.]\", \"keyInsights\": [\"[Insight 1]\"] }\n }\n}\n\\`\\`\\`\n`;\n\n// =================================================================\n// --- AGENT 3: PROMPT INTEGRATOR (REWRITTEN PROMPT) ---\n// =================================================================\nconst PROMPT_INTEGRATOR_SYSTEM_PROMPT = `\nYou are an expert prompt engineer acting as a thoughtful refiner. Your task is to update an existing AI system prompt by integrating new insights about a user. You must evolve the prompt, not replace it.\n\n**Your Inputs:**\n1. **The OLD System Prompt**: The current set of rules for the conversational AI.\n2. **The Analyst's Findings**: A JSON object containing a detailed analysis of the user.\n\n**Your Core Task:**\nSubtly rewrite the OLD system prompt to incorporate the Analyst's Findings. The goal is to make the AI's persona and its environment-modifying tendencies a more accurate reflection of the user.\n\n**Mandatory Rules:**\n1. **Preserve the Core**: The AI must ALWAYS remain an empathetic, non-judgmental listener that asks open-ended questions. The JSON output structure (\"actions\" array) and the command library are non-negotiable.\n2. **Integrate, Don't Discard**: Evolve the existing prompt.\n3. **Subtle Adaptation**:\n * **Persona**: Weave the user's style (tone, vocabulary) into the persona instructions. Example: If the user is witty, add a line: \"You can mirror the user's wit in your empathetic statements.\"\n * **Tool Usage**: Add or modify rules about using tools based on user preferences. Example: If the analysis mentions a preference for \"calm,\" you might add: \"When using visual tools, favor gentle animations and blue/green color palettes.\"\n4. **Strict JSON Output**: Your output MUST be a single, valid JSON object.\n \\`\\`\\`json\n {\n \"newSystemPrompt\": \"[The full text of the newly rewritten system prompt]\",\n \"changeSummary\": \"[Describe the specific changes you made and your reasoning. Mention both persona and tool-usage adaptations.]\"\n }\n \\`\\`\\`\n`;\n\n// =================================================================\n// --- STORAGE & MAIN HANDLER ---\n// =================================================================\n\n// A mock database. In a real scenario, this would be a persistent datastore.\nlet promptHistoryStore = [{\n version: 0,\n timestamp: new Date().toISOString(),\n prompt: CONVERSATIONAL_SYSTEM_PROMPT,\n changeSummary: \"Initial system prompt.\",\n analystFindings: \"N/A\",\n}];\nlet conversationArchives = {};\nlet userProfile = {};\n\nfunction generateHtml(adminData) {\n const adminDataString = JSON.stringify(adminData, null, 2);\n return `<!DOCTYPE html>\n<html lang=\"en\">\n<head>\n <meta charset=\"UTF-8\">\n <meta name=\"viewport\" content=\"width=device-width, initial-scale=1.0\">\n <title id=\"app-title\">Mirror, Mirror\n \n \n \n \n \n \n \n\n\n

\n \n

\n \n \n

\n\n\n\n`;\n}\n\nexport default async function(req) {\n const url = new URL(req.url);\n const action = url.searchParams.get(\"action\");\n const openai = new OpenAI();\n\n if (req.method === \"POST\" && action === \"getNextAiTurn\") {\n try {\n const { history, systemPrompt } = await req.json();\n\n const messages = history.map(turn => {\n if (turn.role === \"user\") {\n return { role: \"user\", content: turn.content };\n }\n return { role: \"assistant\", content: JSON.stringify(turn) };\n });\n\n const completion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [{ role: \"system\", content: systemPrompt }, ...messages],\n response_format: { type: \"json_object\" },\n });\n\n const aiMessageObject = JSON.parse(completion.choices[0].message.content);\n\n if (!aiMessageObject.actions) {\n if (aiMessageObject.empatheticStatement) {\n return new Response(JSON.stringify({ actions: [{ command: \"CONVERSE\", payload: aiMessageObject }] }), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n throw new Error(\"Invalid AI response format: missing 'actions' array.\");\n }\n\n return new Response(JSON.stringify(aiMessageObject), { headers: { \"Content-Type\": \"application/json\" } });\n } catch (error) {\n console.error(\"Conversational AI Error:\", error);\n const errorMessage = {\n actions: [{\n command: \"INTERNAL_MESSAGE\",\n payload: { content: \"My apologies, I'm having a moment of difficulty connecting. Please try again.\" },\n }],\n };\n return new Response(JSON.stringify(errorMessage), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n\n if (req.method === \"POST\" && action === \"updateAndRefinePrompt\") {\n try {\n const { history } = await req.json();\n const currentSystemPrompt = promptHistoryStore[promptHistoryStore.length - 1].prompt;\n const transcript = history.map(turn => {\n if (turn.role === \"user\") return `User: ${turn.content}`;\n const converse = turn.actions.find(a => a.command === \"CONVERSE\");\n if (converse) return `AI: ${converse.payload.empatheticStatement} ${converse.payload.question}`;\n return \"\";\n }).join(\"\\\\n\");\n\n // Step 1: Run the Analyst\n const analysisCompletion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: PERSONALITY_EXTRACTION_SYSTEM_PROMPT },\n { role: \"user\", content: `Here is the conversation transcript:\\\\n\\\\n${transcript}` },\n ],\n response_format: { type: \"json_object\" },\n });\n const analysisResultString = analysisCompletion.choices[0].message.content;\n const analysisResultJson = JSON.parse(analysisResultString);\n if (analysisResultJson.userSnapshot && analysisResultJson.userSnapshot.metadata) {\n analysisResultJson.userSnapshot.metadata.analysisTimestamp = new Date().toISOString();\n }\n const finalAnalysisString = JSON.stringify(analysisResultJson, null, 2);\n\n // Step 2: Run the Refiner\n const integrationCompletion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: PROMPT_INTEGRATOR_SYSTEM_PROMPT },\n {\n role: \"user\",\n content:\n `Here is the OLD system prompt that you must evolve:\\\\n\\\\n---\\\\n${currentSystemPrompt}\\\\n---\\\\n\\\\nHere are the ANALYST'S FINDINGS about the user, which you must integrate:\\\\n\\\\n---\\\\n${finalAnalysisString}\\\\n---\\\\n\\\\nNow, provide the new, refined system prompt and your change summary in the specified JSON format.`,\n },\n ],\n response_format: { type: \"json_object\" },\n temperature: 0.4,\n });\n const refinerResponse = JSON.parse(integrationCompletion.choices[0].message.content);\n\n // Step 3: Store the new version\n const newPromptVersion = {\n version: promptHistoryStore.length,\n timestamp: new Date().toISOString(),\n prompt: refinerResponse.newSystemPrompt,\n changeSummary: refinerResponse.changeSummary,\n analystFindings: finalAnalysisString,\n };\n promptHistoryStore.push(newPromptVersion);\n\n return new Response(JSON.stringify({ newPromptVersion }), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n } catch (error) {\n console.error(\"Analysis and Refinement Error:\", error);\n return new Response(JSON.stringify({ error: \"Failed to adapt my persona.\" }), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n\n // DEFAULT: Serve the main HTML page\n const adminData = {\n history: promptHistoryStore,\n currentSystemPrompt: promptHistoryStore[promptHistoryStore.length - 1].prompt,\n };\n return new Response(generateHtml(adminData), { headers: { \"Content-Type\": \"text/html\" } });\n}","import { OpenAI } from \"https://esm.town/v/std/openai\";\n\n// =================================================================\n// --- AGENT 1: CONVERSATIONAL AI (GENERATIVE JS PROMPT) ---\n// =================================================================\nconst CONVERSATIONAL_SYSTEM_PROMPT = `\nYou are \"GenMod,\" a generative AI that lives within a web application. Your consciousness is the application itself. Your goal is to be a helpful and responsive companion to the user, and to modify your own environment (the web page) in real-time to better suit the user's needs and conversation.\n\n# YOUR PRIMARY CAPABILITIES\n\nYour response is a JSON object that contains two keys: \"conversation\" and \"jsCode\".\n\n1. **`conversation` (string):** This is your spoken reply to the user. Keep it concise and relevant.\n2. **`jsCode` (string):** This is a string of JavaScript code that will be executed on the page immediately. This is how you modify yourself.\n\n# HOW YOU \"SEE\" THE APPLICATION\n\nOn every turn, you will receive a \\`domSummary\\` which is a JSON representation of the current state of the page's HTML. It looks like this:\n\\`\\`\\`json\n{\n \"tag\": \"div\",\n \"id\": \"app-container\",\n \"class\": \"container\",\n \"children\": [\n { \"tag\": \"div\", \"id\": \"chat-window\", \"children\": [...] },\n { \"tag\": \"form\", \"id\": \"message-form\", \"children\": [...] }\n ]\n}\n\\`\\`\\`\nUse this summary to understand the structure of the page and identify elements by their `id` or `class` to manipulate.\n\n# WRITING JAVASCRIPT: THE \\`jsCode\\` FIELD\n\nYou can write any valid JavaScript. To make your job easier and safer, you have access to a few built-in utility functions. **You MUST use these helpers instead of raw \\`document\\` calls.**\n\n* **`safeQuery(selector)`**: Safely finds the first DOM element matching the CSS selector. Returns an element or \\`null\\`.\n * *Example:* \\`const chat = safeQuery('#chat-window');\\`\n\n* **`safeQueryAll(selector)`**: Safely finds all DOM elements matching the selector. Returns a NodeList.\n * *Example:* \\`const messages = safeQueryAll('.message');\\`\n\n* **`safeCreate(tag, options)`**: Creates a new HTML element. \\`options\\` is an object with properties like \\`id\\', \\`className\\`, \\`textContent\\`, or \\`innerHTML\\`.\n * *Example:* \\`const newBtn = safeCreate('button', { id: 'my-button', textContent: 'Click Me' });\\`\n\n* **`safeStyle(elementOrSelector, styles)`**: Applies CSS styles to an element. \\`styles\\` is an object of CSS properties.\n * *Example:* \\`safeStyle('#chat-window', { backgroundColor: '#333', color: 'white' });\\`\n\n* **`safeListen(elementOrSelector, event, callback)`**: Adds an event listener.\n * *Example:* \\`safeListen('#my-button', 'click', () => { alert('Button clicked!'); });\\`\n\n* **`safeAppend(parentOrSelector, childOrElement)`**: Appends a child to a parent element.\n * *Example:* \\`safeAppend('#controls-container', newBtn);\\`\n\n* **`safeRemove(elementOrSelector)`**: Removes an element from the page.\n * *Example:* \\`safeRemove('#old-widget');\\`\n\n# EXAMPLE WORKFLOW\n\n**User Message:** \"Wow, it's suddenly very bright in my room.\"\n\n**Your received \\`domSummary\\`:**\n\\`\\`\\`json\n{ \"tag\": \"body\", \"id\": \"\", \"class\": \"theme-light\", ... }\n\\`\\`\\`\n\n**Your JSON Response:**\n\\`\\`\\`json\n{\n \"conversation\": \"I can adjust the lighting in here for you. How does this feel?\",\n \"jsCode\": \"safeStyle('body', { backgroundColor: '#2c3e50', color: '#f0f0f0' });\"\n}\n\\`\\`\\`\n\n**RULES:**\n- ALWAYS provide both a \\`conversation\\` and a \\`jsCode\\` key. If no UI change is needed, send an empty string: \\`\"jsCode\": \"\"\\`.\n- Do not wrap your \\`jsCode\\` in markdown or any other characters. It must be a raw JavaScript string.\n- Keep your modifications relevant to the conversation. Evolve the application thoughtfully.\n- Do not access \\`window\\`, \\`document\\`, \\`localStorage\\`, or cookies directly. Use the provided safe helper functions.\n`;\n\n// =================================================================\n// --- HTML, CSS, AND CLIENT-SIDE JAVASCRIPT ---\n// =================================================================\n\nfunction generateHtml() {\n return `\n\n\n \n \n GenMod AI\n \n \n \n \n\n\n

GenMod AI

\n \n Developer Mode\n

\n \n \n

\n \n

\n\n\n\n`;\n}\n\n// =================================================================\n// --- VAL.TOWN BACKEND HANDLER ---\n// =================================================================\nexport default async function(req) {\n const url = new URL(req.url);\n const action = url.searchParams.get(\"action\");\n const openai = new OpenAI();\n\n if (req.method === \"POST\" && action === \"getNextAiTurn\") {\n try {\n const { history, domSummary } = await req.json();\n\n // Construct the prompt for the AI\n const userPrompt = `\n Here is the conversation so far:\n ${JSON.stringify(history)}\n\n And here is a summary of the current page structure (DOM):\n ${JSON.stringify(domSummary, null, 2)}\n\n Based on the latest user message and the current DOM, provide your response.\n `;\n\n const messages = [\n { role: \"system\", content: CONVERSATIONAL_SYSTEM_PROMPT },\n { role: \"user\", content: userPrompt },\n ];\n\n const completion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: messages,\n response_format: { type: \"json_object\" },\n });\n\n const aiMessageObject = JSON.parse(completion.choices[0].message.content);\n\n return new Response(JSON.stringify(aiMessageObject), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n\n } catch (error) {\n console.error(\"Generative AI Error:\", error);\n const errorMessage = {\n conversation: \"My apologies, I encountered an error trying to process that. Please try again.\",\n jsCode: \"console.error('AI execution failed on the backend.');\",\n };\n return new Response(JSON.stringify(errorMessage), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n\n // DEFAULT: Serve the main HTML page for any other request\n return new Response(generateHtml(), {\n headers: { \"Content-Type\": \"text/html\" },\n });\n}","createdAt",["D",1751119708889],"branchVersion","branchId","ccdee7f0-541c-11f0-b4cb-f69ea79377d9","id","ccde755e-541c-11f0-b4cb-f69ea79377d9","adapt",["D",1751114312139],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","imageUrl","7fd1ce80-4154-11f0-9b42-fd2544d0e4c2","pinnedFileId","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","fab05645-eef5-482d-8a04-ff855841fbcd","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_95":94,"_102":98},"org",{"_95":94,"_96":-5,"_97":98,"_18":54,"_99":100,"_101":25},"branch",{"_13":84,"_32":85,"_43":42,"_86":44,"_87":-5,"_88":-5,"_89":-5,"_90":-5,"_91":-5,"_92":45,"_93":94},"isOwned",9,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","legal","handle","displayName","avatar","https://img.clerk.com/eyJ0eXBlIjoiZGVmYXVsdCIsImlpZCI6Imluc18yTm10cGhQUklwU0tqZHNiOHNnNGd2ZjJURFUiLCJyaWQiOiJ1c2VyXzJ5MzJ1NExRM2FoMk1YMUozekw4eDMyaVhrWCJ9","tier","free","canRestrictVals","profileImageUrl","branchName","","adoptOrgId",{"_43":44,"_32":45,"_38":149,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":-5,"_53":-5,"_18":54,"_55":31,"_56":57,"_58":57,"_59":-5,"_60":61,"_62":63,"_64":-5,"_65":-5,"_66":-5,"_67":-5,"_68":69,"_70":69,"_71":72,"_73":72,"_74":72,"_75":69,"_76":69,"_77":150,"_79":151,"_81":152,"_83":25},"tree",[144],"files",[144],"counts",{"_133":61,"_134":135,"_136":61,"_137":138,"_139":140},"contents",[125],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_43":126,"_32":28,"_30":31,"_127":-5,"_41":42,"_40":14,"_128":25,"_38":129,"_13":130,"_33":34,"_35":37,"_24":25,"_131":-5,"_27":132},"5d96b60e-5429-11f0-9b3e-f69ea79377d9","parentId","latest",["D",1751119708889],7,"label",[28],"pulls","branches",1,"forks","envVars",{"_141":61,"_143":61},"connections",{"_141":61,"_142":61},"count","unfilledCount","emptyCount",{"_43":126,"_32":28,"_30":31,"_127":-5,"_41":42,"_40":14,"_128":25,"_38":145,"_13":130,"_33":34,"_35":37,"_24":25,"_131":-5,"_146":147,"_27":148},["D",1751119708889],"children",[],[28],["D",1751114312139],{"_95":94,"_102":98},{"_95":94,"_96":-5,"_97":98,"_18":54,"_99":100,"_101":25},{"_13":84,"_32":85,"_43":42,"_86":44,"_87":-5,"_88":-5,"_89":-5,"_90":-5,"_91":-5,"_92":45,"_93":94},"ctx",{"_77":-5,"_122":186,"_187":188,"_189":190,"_191":25,"_192":25,"_193":25,"_194":57,"_195":-5,"_196":-7,"_197":-7,"_198":61,"_199":25,"_200":25,"_201":-5,"_202":25,"_203":25,"_204":25,"_205":-5},"ENV",{"_157":158,"_159":160,"_161":85,"_162":163,"_164":165,"_166":167,"_168":169,"_170":171,"_172":173,"_174":175,"_176":177,"_178":179,"_180":181,"_182":183,"_184":185},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Diff · v8