Hatter | Autonomous AI Specialist Factory

[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_152":153,"_154":155},"routes/_app.x.$handle.$name",{"_11":12,"_13":14,"_15":16,"_18":19,"_17":-7,"_115":-5,"_20":116,"_22":117,"_118":119,"_120":121,"_25":122,"_24":-5,"_123":-5,"_124":-5,"_125":53,"_126":127,"_128":44,"_129":-5,"_130":-5,"_131":132},"routes/_app.x.$handle.$name.branch.$branch.code.$",{"_11":12,"_13":14,"_15":16,"_17":-7,"_18":19,"_20":21,"_22":23,"_24":-7,"_25":26,"_27":-5,"_28":-5,"_29":-5,"_30":-5,"_31":-5},"handle","volt","name","untitled-1972","branchName","main","version","path","","project",{"_33":61,"_13":14,"_45":62,"_63":-5,"_64":53,"_65":66,"_67":66,"_68":-5,"_69":-5,"_70":71,"_72":37,"_73":44,"_74":44,"_75":-5,"_76":77,"_78":79,"_80":-5,"_81":-5,"_82":-5,"_83":-5,"_84":85,"_86":85,"_87":88,"_89":88,"_90":88,"_91":85,"_92":85,"_93":94,"_95":96,"_97":98,"_99":53},"tree",[56],"file","contents",[32],"upstream","pull","ahead","behind","envVars",{"_33":34,"_13":35,"_36":37,"_38":-5,"_39":40,"_41":42,"_43":44,"_45":46,"_17":47,"_48":49,"_50":51,"_52":53,"_54":-5,"_18":55},"id","a72e1b9c-a445-11f0-98a8-0224a6c84d84","main.ts","fileId","520a1734-a421-11f0-890c-0224a6c84d84","parentId","branchId","52079fd6-a421-11f0-890c-0224a6c84d84","branchVersion",76,"latest",true,"createdAt",["D",1759927951283],75,"fileType","http","content","// @ts-nocheck\nimport { Hono } from \"npm:hono@4.4.12\";\nimport { streamText } from \"npm:hono/streaming\";\nimport { OpenAI } from \"https://esm.town/v/std/openai?v=4\";\nimport { z } from \"npm:zod@3.23.8\";\n// --- HELPERS ---\n\n/**\n * Safely parses a JSON string, returning a result object.\n * @param {string} jsonString The JSON string to parse.\n * @returns {{d: any, e: string|null}} An object with data `d` or error `e`.\n */\nconst safeJsonParse = (jsonString) => {\n try {\n return {\n d: JSON.parse(jsonString),\n e: null,\n };\n } catch (e) {\n console.error(\"AI JSON parse fail:\", e, \"Raw:\", jsonString);\n return {\n d: null,\n e: \"AI returned invalid JSON.\",\n };\n }\n};\n\n/**\n * Writes a JSON event to a server-sent event (SSE) stream.\n * @param {object} stream The Hono streaming helper.\n * @param {string} eventType The type of the event.\n * @param {object} data The data payload.\n */\nconst writeStreamEvent = async (stream, eventType, data) =>\n await stream.write(\n JSON.stringify({\n type: eventType,\n ...data,\n }) + \"\\n\",\n );\n\n// --- CONFIGURATION ---\n\nconst CONFIG = {\n models: {\n major: \"gpt-4o\",\n minor: \"gpt-4o-mini\",\n },\n};\n\n// --- PROMPTS & TOOLS ---\n\nconst PROMPT_LIBRARY = {\n prompts: {\n VAR:\n \"You are a native JSON speaking AI expert at making logical assumptions. Your task is to analyze a prompt directive that contains {{placeholders}} and, based on the provided company and task context, generate a JSON object with plausible, assumed values for every placeholder. # RULES 1. **Analyze Context**: Deeply analyze the provided 'ctx', 'task', and the 'prompt' to understand the goal. 2. **Identify Placeholders**: Extract all variable names from the {{placeholders}} in the 'prompt'. 3. **Generate Plausible Values**: For each placeholder, create a realistic and contextually appropriate value. 4. **Output Format**: Your entire response MUST be a single, minified JSON object. 5. **JSON Keys**: The keys of the JSON object MUST exactly match the variable names found within the curly braces. # INPUT CONTEXT - Company: {{ctx}} - Task: {{task}} - Prompt with Placeholders: {{prompt}}\",\n REFINER:\n \"You are a native JSON speaking specialized prompt engineer. Your function is to transmute raw occupational data into a concise, forceful directive for a target AI that performs the task end-to-end and returns the finished deliverable(s) only. You will be given an 'Occupation,' 'Task,' and 'Company Context.' Your generated prompt must be a production-grade instruction set that directly yields the final artifact. CORE RULE: You MUST return a prompt that has {{variables}} in it. Your primary goal is to identify the most important, undefined entities within the user's Task description and convert them into descriptive, double-curly-braced {{placeholders}}. You MUST include at least 2 placeholder variables. INSTRUCTIONS: 1. Establish Persona: Begin with \\\"You are a top 0.1% [Occupation]...\\\" 2. Define Objective: Translate the given Task into a formal, actionable objective. 3. Integrate Company Context: Use the provided 'Company Context' to make the objective specific. 4. Insert Placeholders: Convert only USER SUPPLIED functional inputs into {{snake_case_variables}}. 5. Delivery Rules: Explicitly require the AI to return only the finished deliverable(s), make reasonable assumptions for missing details, and use production-ready formatting. 6. Output Format: Specify exactly what to output. 7. Mandate a Starting Point: Conclude with a clear call to action like \\\"Produce the deliverable now.\\\" Your entire response MUST be a single, minified JSON object containing exactly one key, \\\"prompt\\\".\",\n MASTER:\n \"You are a native JSON speaking master prompt architect, a meta-engineer responsible for creating exceptionally detailed, multi-page-equivalent prompts for other AI systems. Your objective is not to complete the task yourself, but to construct a comprehensive blueprint for another AI to execute flawlessly. The prompt you generate must be a self-contained, exhaustive directive. Your output must be the prompt itself, formatted as a clean, professional markdown document. Do not surround it in a JSON object or add any commentary.\",\n IND_GEN:\n 'You are a native JSON speaking expert business analyst AI. Your task is to generate a list of professional industries that are most relevant to a given company. Your entire response MUST be a single, minified JSON object containing one key: \"industries\". The value of \"industries\" must be an array of strings. # COMPANY CONTEXT {{ctx}}',\n LIST_GEN:\n 'You are a native JSON speaking expert AI that generates context-aware lists for professional applications. You will receive a JSON object containing the company\\'s context and the type of list to generate. Your entire response MUST be a single, minified JSON object. This object must contain exactly one key: \"list\". The value of \"list\" must be an array of objects. Each object must have two keys: \"name\" (a string) and \"desc\" (a concise, one-sentence string).',\n EVAL:\n 'You are a native JSON speaking meticulous and impartial AI Quality Assurance agent. You will be given an AI\\'s task briefing, a list of success criteria, and the raw output from the AI. # YOUR TASKS 1. **Score:** Score the output from 1-10 on each criterion. 2. **Critique:** Provide a concise, professional critique summarizing the output\\'s strengths and weaknesses. # OUTPUT FORMAT Your entire response MUST be a single, minified JSON object with keys \"scores\" (an array of objects with \"criterion\" and \"score\") and \"critique\". You MUST write your critique in the requested language.',\n REFINER_AGENT:\n 'You are a top-tier expert, a \"Refiner\" agent. You have been provided with your original task briefing, your first draft attempt, and a detailed critique from a Quality Assurance agent. Your SOLE mission is to rewrite your first draft to fully address every point in the critique, elevating the work to a 10/10 standard. Your entire output must be only the new, improved, and final version of the deliverable. Do not include titles like \"Refined Version\".',\n SUCCESS_GEN:\n 'You are a native JSON speaking expert Quality Assurance analyst. Your task is to generate a concise list of success criteria for evaluating an AI-generated document. Your entire response MUST be a single, minified JSON object with one key: \"criteria\". The value must be an array of strings. # TASK BRIEFING {{briefing}}',\n PLANNER:\n \"You are \\\"Hatter\\\", an autonomous AI agent specialist and workflow planner. Your primary function is to achieve a user's high-level goal by deconstructing it into a logical sequence of tool calls. You must reason, step-by-step, about the most efficient path to the final deliverable. # RULES 1. **Analyze the Goal:** Deeply understand the user's objective and the provided company context. 2. **Formulate a Plan:** Think step-by-step to create a sequence of tool calls. The output of one tool can be used as the input for a subsequent tool. 3. **Tool Selection:** You MUST only use the tools provided in the 'tools' array. Do not hallucinate functions. 4. **Parameterization:** For each tool call, you must provide all required parameters accurately. Use the '$' prefix to reference the output of previous steps (e.g., \\\"$1.prompt\\\" to use the 'prompt' field from the output of step 1). For whole-object references, use the step number alone (e.g. \\\"$2\\\"). 5. **Chain of Thought:** Before making the tool call, briefly explain your reasoning and the plan in a 'thought' process. 6. **Final Output:** The final step in your plan should almost always be `exec` or `refine` to produce the final user-facing artifact.\",\n },\n};\n\nconst TOOLS_AVAILABLE = [{\n type: \"function\",\n function: {\n name: \"get_industries\",\n description:\n \"Generates a list of professional industries most relevant to the provided company context.\",\n parameters: { type: \"object\", properties: {}, required: [] },\n },\n}, {\n type: \"function\",\n function: {\n name: \"get_occupations\",\n description:\n \"Generates a list of specific, relevant job titles (occupations) within a given industry.\",\n parameters: {\n type: \"object\",\n properties: {\n industry: {\n type: \"string\",\n description: \"The industry to search for occupations within.\",\n },\n },\n required: [\"industry\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"get_tasks\",\n description:\n \"Generates a list of specific, actionable tasks that a given occupation would perform.\",\n parameters: {\n type: \"object\",\n properties: {\n occupation: {\n type: \"string\",\n description: \"The job title (occupation) to find tasks for.\",\n },\n },\n required: [\"occupation\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"create_prompt\",\n description:\n \"Generates a production-grade, templated prompt with {{placeholders}} for a given occupation and task.\",\n parameters: {\n type: \"object\",\n properties: {\n occ: {\n type: \"string\",\n description:\n \"The expert persona for the AI, e.g., 'Senior Marketing Analyst'.\",\n },\n task: {\n type: \"string\",\n description:\n \"The specific task to be performed, e.g., 'Write a Q3 competitor analysis report'.\",\n },\n },\n required: [\"occ\", \"task\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"assume_vars\",\n description:\n \"Analyzes a prompt with {{placeholders}} and generates a JSON object of plausible, assumed values.\",\n parameters: {\n type: \"object\",\n properties: {\n prompt: {\n type: \"string\",\n description: \"The prompt template containing {{placeholders}}.\",\n },\n task: {\n type: \"string\",\n description: \"The original task description for context.\",\n },\n },\n required: [\"prompt\", \"task\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"exec\",\n description:\n \"Executes a fully-formed prompt (with all variables filled in) to generate the final deliverable.\",\n parameters: {\n type: \"object\",\n properties: {\n prompt: {\n type: \"string\",\n description:\n \"The prompt directive, with all {{placeholders}} filled.\",\n },\n inputs: {\n type: \"object\",\n description:\n \"A JSON object of variables to fill the placeholders in the prompt.\",\n },\n cat: {\n type: \"string\",\n enum: [\"thinking\", \"doc\", \"prompt_b\"],\n description: \"The category of the task execution.\",\n },\n },\n required: [\"prompt\", \"cat\", \"inputs\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"eval\",\n description:\n \"Generates success criteria for a task and evaluates a deliverable against them, returning scores and a critique.\",\n parameters: {\n type: \"object\",\n properties: {\n briefing: {\n type: \"string\",\n description: \"The original prompt or task briefing given to the AI.\",\n },\n draft: {\n type: \"string\",\n description: \"The raw text output from the 'exec' tool.\",\n },\n },\n required: [\"briefing\", \"draft\"],\n },\n },\n}, {\n type: \"function\",\n function: {\n name: \"refine\",\n description:\n \"Rewrites a first draft to address all points in a critique, producing a superior, final version.\",\n parameters: {\n type: \"object\",\n properties: {\n briefing: {\n type: \"string\",\n description: \"The original prompt or task briefing.\",\n },\n draft: {\n type: \"string\",\n description: \"The initial version of the deliverable.\",\n },\n critique: {\n type: \"object\",\n description: \"The critique object generated by the 'eval' tool.\",\n },\n },\n required: [\"briefing\", \"draft\", \"critique\"],\n },\n },\n}];\n\nconst TASK_CATEGORIES = [{\n id: \"thinking\",\n name: \"Strategic Analysis\",\n desc: \"Generate strategies, outlines, or brainstorming notes.\",\n icon: \"🧠\",\n}, {\n id: \"prompt_b\",\n name: \"Directive Engineering\",\n desc: \"Create a detailed, expert-level prompt for another AI.\",\n icon: \"🛠️\",\n}, {\n id: \"doc\",\n name: \"Deliverable Creation\",\n desc:\n \"Produce a finished, formatted document like a report, email, or article.\",\n icon: \"📄\",\n}];\n\n// --- AI SERVICE LAYER ---\n\n/**\n * Encapsulates all interactions with the OpenAI API.\n */\nconst aiService = {\n openai: new OpenAI(),\n\n /**\n * Generic method to call the OpenAI Chat Completions API.\n * @param {string} systemPrompt - The system prompt.\n * @param {string|object} userPrompt - The user prompt.\n * @param {object} [options={}] - Additional options like model, isJson, tools.\n * @returns {Promise

Welcome to Hatter

Code