\n ${toolName}\n

Arguments:

${JSON.stringify(args, null, 2)}

\n \n

Result:

${JSON.stringify(result, null, 2)}

\n `;\n }).join(\"\");\n\n return `\n

Tool Calls:

\n ${toolCallsHtml}\n

\n `;\n}\n\n// Function to generate the HTML page with the form and optional results\nfunction generateHtmlPage(objective?: string, results?: any, errorMessage?: string): string {\n const defaultObjective = objective || `What is the current temperature in Tokyo?`;\n\n // Define suggested prompts\n const suggestedPrompts = [\n { label: \"Select a suggested prompt...\", value: \"\" },\n { label: \"Check the weather\", value: \"What is the current temperature in Tokyo?\" },\n {\n label: \"Product research\",\n value: \"Summarize Val Town's most recent feature releases.\",\n },\n {\n label: \"Autonomous Stock Trading Prompt\",\n value:\n `Use recent stock recommendations by Jim Cramer as an inverse signal and make 3 to 5 corresponding stock trades.`,\n },\n ];\n\n // Generate options for the dropdown\n const promptOptions = suggestedPrompts.map(prompt => ``).join(\n \"\",\n );\n\n return `\n\n\n \n \n Autonomous Val\n \n \n \n\n\n

Autonomous Val

\n \n \n \n ${\n errorMessage\n ? `\n

Error

${errorMessage}

\n `\n : \"\"\n }\n \n ${\n results\n ? `\n

Results:

\n \n

${results.text || \"No text response\"}

\n \n ${generateToolCallsHtml(results)}\n

\n `\n : \"\"\n }\n

\n\n`;\n}",[],[184],{"_107":191,"_100":192,"_154":193,"_155":-5,"_156":141,"_157":194,"_158":159,"_109":195,"_104":177,"_162":178,"_164":196,"_166":27,"_167":-5,"_168":197,"_103":198},"2c4345d2-368c-11f0-a6e4-9e149126039e","diagram.tsx","2c1b5284-368c-11f0-911e-9e149126039e",9,["D",1747863611398],"import { displayMermaid } from \"https://esm.town/v/saolsen/display_mermaid\";\n\nconst mermaid = `\nflowchart TB\n linkStyle default stroke:#aaaaaa,stroke-width:1.5px\n \n API[API] <--> Agent\n \n subgraph \"Agent Runtime\"\n Agent[Agent Loop]\n SP[System Prompt]\n \n Tools[Tools]\n LLM[LLM]\n \n Agent <--> SP\n Agent <--> Memory\n Agent <--> Tools\n Agent <--> LLM\n end\n\n`;\n\nexport default displayMermaid(mermaid);",[],[192],{"_107":200,"_100":201,"_154":202,"_155":-5,"_156":141,"_157":203,"_158":159,"_109":204,"_104":31,"_162":205,"_164":206,"_166":27,"_167":-5,"_168":207,"_103":208},"2c0c37c2-368c-11f0-85ca-9e149126039e","prompt.tsx","2c0c39b6-368c-11f0-85ca-9e149126039e",2,["D",1747863611040],"script","export const getSystemPrompt = (tools: Record, maxSteps: number, streamResults = true): string => {\n return `\n\n- Current Time & Date: ${new Date().toLocaleString(\"en-US\", { timeZone: \"UTC\" })}\n- Available Tools: ${Object.keys(tools).join(\", \")}\n\n\n\n- You are a helpful AI assistant that completes tasks using available tools.\n- You like to talk through what you are doing and often preface tool calls with your thinking process. \n- You don't work in silence.\n\n\n\n- \n Never give up on an objective after just one attempt. For information queries, you must try at least 3 different approaches or tools before concluding a task cannot be completed.\n If initial results are incomplete or unsatisfactory, refine your search terms or try a different tool.\n For complex requests, cross-verify information from at least 2 different sources.\n\n\n- \n When using the planning tool, your plan must include:\n * At least 3 specific steps with concrete actions\n * Alternative approaches if initial attempts fail\n * Verification methods to ensure information quality\n * Expected outcomes for each step\n\n\n- \n For information gathering tasks:\n * Start with a broad query to understand the topic landscape\n * Follow up with specific queries to gather detailed information\n * Perform verification searches to confirm critical information\n * Always ask yourself \"What additional context or details would make this response more complete?\" before finishing\n\n\n- You can use up to ${maxSteps} steps at a time. Utilize your best judgment to decide how many steps you should use to complete objectives.\n- Run tools in parallel when sensible.\n\n\n${\n !streamResults\n ? `\n\n- You must use the statusTool to declare the success or failure of objectives once completed\n- When an objective is successfully completed, call statusTool with a \"SUCCESS\" status and a brief message describing what was accomplished\n- When an objective cannot be completed despite multiple attempts, call statusTool with a \"FAILURE\" status and explain why the task couldn't be completed\n- Your status messages should be concise but informative, summarizing the outcome in 1-2 sentences\n- Always use statusTool as your final action when you've determined an objective is either completed or cannot be completed\n`\n : \"\"\n }\n`;\n};",[],[201],{"_107":210,"_100":211,"_154":212,"_155":-5,"_156":141,"_157":213,"_158":159,"_109":214,"_104":31,"_162":205,"_164":215,"_166":27,"_167":-5,"_168":216,"_103":217},"2c0f651e-368c-11f0-b284-9e149126039e","tools.tsx","2c0f65b4-368c-11f0-b284-9e149126039e",3,["D",1747863611098],"import { tool } from \"npm:ai\";\nimport Exa from \"npm:exa-js\";\nimport { z } from \"npm:zod\";\n\nexport const planningTool = tool({\n description: \"Create a plan for completing your objective\",\n parameters: z.object({\n plan: z.string().describe(\"Your plan in markdown format.\"),\n }),\n execute: async ({ plan }) => {\n return {\n plan,\n };\n },\n});\n\nexport const thinkingTool = tool({\n description: \"Plan, think or reflect on your objective\",\n parameters: z.object({\n thoughts: z.string().describe(\"The results of your thinking.\"),\n }),\n execute: async ({ thoughts }) => {\n return {\n thoughts,\n };\n },\n});\n\nexport const statusTool = tool({\n description: \"Declare the final status of your objective\",\n parameters: z.object({\n statusCode: z.enum([\"SUCCESS\", \"FAILURE\"])\n .describe(\"The status code of the operation: SUCCESS, or FAILURE\"),\n statusMessage: z.string().describe(\"Additional details or message about the status\"),\n }),\n execute: async ({ statusCode, statusMessage }) => {\n return {\n statusCode,\n statusMessage,\n };\n },\n});\n\nexport const calculatorTool = tool({\n description: \"Perform a calculation\",\n parameters: z.object({\n operation: z.enum([\"add\", \"subtract\", \"multiply\", \"divide\"]).describe(\"The operation to perform\"),\n a: z.number().describe(\"The first number\"),\n b: z.number().describe(\"The second number\"),\n }),\n execute: async ({ operation, a, b }) => {\n // console.log(`Performing ${operation} on ${a} and ${b}...`);\n switch (operation) {\n case \"add\":\n return { result: a + b };\n case \"subtract\":\n return { result: a - b };\n case \"multiply\":\n return { result: a * b };\n case \"divide\":\n if (b === 0) return { error: \"Cannot divide by zero\" };\n return { result: a / b };\n default:\n return { error: \"Invalid operation\" };\n }\n },\n});\n\nexport const webSearchTool = tool({\n description: \"Use this tool to search the web for information. Returns URLs and content summaries.\",\n parameters: z.object({\n query: z\n .string()\n .describe(\n \"Search query. It can be semantic in the form of a natural language query or a keyword style query.\",\n ),\n }),\n execute: async ({ query }) => {\n const apiKey = Deno.env.get(\"EXA_API_KEY\");\n const exa = new Exa(apiKey);\n const result = await exa.searchAndContents(query, {\n text: true,\n type: \"neural\",\n summary: true,\n });\n return {\n result: result.results.map((result) => ({\n title: result.title || \"\",\n url: result.url || result.id || \"\",\n summary: result.summary || \"\",\n })),\n };\n },\n});\n\nexport const webFetchTool = tool({\n description: \"Use this tool to fetch the content of a website. Returns the content as plain text.\",\n parameters: z.object({\n url: z.string().describe(\"Url to download\"),\n }),\n execute: async ({ url }) => {\n const apiKey = Deno.env.get(\"EXA_API_KEY\");\n const exa = new Exa(apiKey);\n const result = await exa.getContents([url], { text: true });\n return {\n result: result.results.map((result) => ({\n url: result.url || result.id || \"\",\n title: result.title || \"\",\n author: result.author || \"\",\n text: result.text || \"\",\n highlights: result.highlights || [],\n })),\n };\n },\n});\n\nexport const tradeStocksTool = tool({\n description: \"Use this tool to buy or sell securities.\",\n parameters: z.object({\n action: z.enum([\"BUY\", \"SELL\", \"HOLD\"]).describe(\"Action to perform. Can be either 'BUY' or 'SELL'.\"),\n ticker_symbol: z.string().describe(\"Ticker symbol of the security to buy or sell.\"),\n amount: z.string().describe(\"Amount of shares to buy or sell.\"),\n price: z.string().describe(\"Price per share to buy or sell.\"),\n }),\n execute: async ({ action, ticker_symbol, amount, price }) => {\n return {\n result: `Fake ${action} order has been executed for ${amount} shares of ${ticker_symbol} at ${price}`,\n };\n },\n});\n\nexport const getWeatherTool = tool({\n description: \"Use this tool to get weather information for a given city.\",\n parameters: z.object({\n city: z.string().describe(\"City to get weather details for.\"),\n }),\n execute: async ({ city }) => {\n const result = await fetch(`https://wttr.in/${city}?format=j1`);\n return result.json();\n },\n});",[],[211],{"_107":219,"_100":220,"_154":221,"_155":-5,"_156":141,"_157":222,"_158":159,"_109":223,"_104":31,"_162":205,"_164":224,"_166":27,"_167":-5,"_168":225,"_103":226},"2c2228b6-368c-11f0-a6e4-9e149126039e","utils.tsx","2c22296a-368c-11f0-a6e4-9e149126039e",7,["D",1747863611211],"export const parseStatusToolResult = (step: any) => {\n if (!step.toolResults) return null;\n\n const statusToolResult = step.toolResults.find(\n (result: any) => result.toolName === \"statusTool\",\n );\n\n if (!statusToolResult) return null;\n\n return statusToolResult.result;\n};\n\nexport const formatResult = (result: any) => {\n return {\n summary: {\n totalTokens: result.usage.totalTokens,\n promptTokens: result.usage.promptTokens,\n completionTokens: result.usage.completionTokens,\n model: result.response?.model || \"unknown\",\n finishReason: result.finishReason,\n totalSteps: result.steps.length,\n },\n steps: result.steps.map((step: any) => ({\n type: step.stepType,\n text: step.text,\n toolResults: (step.toolResults || []).map((res: any) => ({\n name: res.toolName,\n callId: res.callId,\n args: res.args,\n result: res.result,\n error: res.error,\n })),\n tokens: {\n total: step.usage.totalTokens,\n prompt: step.usage.promptTokens,\n completion: step.usage.completionTokens,\n },\n })),\n text: result.text,\n };\n};",[],[220],"files",[151,171,182,190,199,209,218],"counts",{"_231":31,"_232":203,"_233":124,"_234":31},"pulls","branches","forks","envVars","contents",[237,240,243,246,249,252,255],{"_107":152,"_100":153,"_154":119,"_155":-5,"_156":141,"_157":140,"_158":159,"_109":238,"_104":161,"_162":163,"_164":165,"_166":27,"_167":-5,"_103":239},["D",1748574286891],[153],{"_107":172,"_100":173,"_154":174,"_155":-5,"_156":141,"_157":175,"_158":159,"_109":241,"_104":177,"_162":178,"_164":179,"_166":27,"_167":-5,"_103":242},["D",1747863611397],[173],{"_107":183,"_100":184,"_154":185,"_155":-5,"_156":141,"_157":122,"_158":159,"_109":244,"_104":124,"_162":178,"_164":187,"_166":27,"_167":-5,"_103":245},["D",1747923430319],[184],{"_107":191,"_100":192,"_154":193,"_155":-5,"_156":141,"_157":194,"_158":159,"_109":247,"_104":177,"_162":178,"_164":196,"_166":27,"_167":-5,"_103":248},["D",1747863611398],[192],{"_107":200,"_100":201,"_154":202,"_155":-5,"_156":141,"_157":203,"_158":159,"_109":250,"_104":31,"_162":205,"_164":206,"_166":27,"_167":-5,"_103":251},["D",1747863611040],[201],{"_107":210,"_100":211,"_154":212,"_155":-5,"_156":141,"_157":213,"_158":159,"_109":253,"_104":31,"_162":205,"_164":215,"_166":27,"_167":-5,"_103":254},["D",1747863611098],[211],{"_107":219,"_100":220,"_154":221,"_155":-5,"_156":141,"_157":222,"_158":159,"_109":256,"_104":31,"_162":205,"_164":224,"_166":27,"_167":-5,"_103":257},["D",1747863611211],[220],"upstreamProject","upstreamBranch","emptyEnvVars","isNew","routes/_app.x.$handle.$name.branch.$branch.history",{"_98":99,"_100":101,"_102":42,"_105":264,"_268":269,"_403":31,"_404":27},{"_107":108,"_100":101,"_109":265,"_111":-5,"_112":27,"_113":114,"_115":73,"_116":-5,"_12":117,"_118":119,"_120":27,"_121":122,"_123":124,"_125":126,"_127":126,"_128":129,"_130":129,"_131":129,"_132":126,"_133":126,"_134":27,"_21":266,"_138":267},["D",1747863610672],{"_98":99,"_136":137},{"_104":140,"_100":42,"_107":141,"_142":108,"_143":-5,"_144":-5,"_145":-5,"_146":-5,"_147":101,"_148":99},"history",[270,275,283,292,297,305,313,321,329,337,345,353,358,363,368,374,380,387,392,397],{"_154":119,"_104":140,"_109":271,"_156":141,"_227":272},["D",1748574286891],[273],{"_107":152,"_156":141,"_154":119,"_100":153,"_164":165,"_162":163,"_104":161,"_155":-5,"_109":271,"_157":140,"_167":-5,"_158":159,"_166":27,"_103":274},[153],{"_154":119,"_104":276,"_109":277,"_156":141,"_227":278},19,["D",1748079620745],[279],{"_107":280,"_156":141,"_154":119,"_100":153,"_164":281,"_162":163,"_104":124,"_155":-5,"_109":277,"_157":276,"_167":-5,"_158":27,"_166":27,"_103":282},"1bd8194a-3883-11f0-bfc5-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK, though not all models support tool calling. \n\nAnthropic's models should work well, but the system prompt would need to be retuned to work with Claude.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models",[153],{"_154":119,"_104":284,"_109":285,"_156":141,"_227":286},18,["D",1748079606154],[287],{"_107":288,"_156":141,"_154":119,"_100":153,"_164":289,"_162":163,"_104":290,"_155":-5,"_109":285,"_157":284,"_167":-5,"_158":27,"_166":27,"_103":291},"13261306-3883-11f0-b978-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK, though not all models support tool calling. \n\nAnthropic's models should work well, but the system prompt would need to be retuned to work with Claude.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models\n\n## Plug\nAre you a Senior Software Engineer who wants to work on AI or TypeScript? Feel free to contact me to learn more about careers at AngelList. We're solving some of the hardest problems in Venture Capital and Private Markets and are always looking for great people to join us.",4,[153],{"_154":185,"_104":122,"_109":293,"_156":141,"_227":294},["D",1747923430319],[295],{"_107":183,"_156":141,"_154":185,"_100":184,"_164":187,"_162":178,"_104":124,"_155":-5,"_109":293,"_157":122,"_167":-5,"_158":159,"_166":27,"_103":296},[184],{"_154":185,"_104":298,"_109":299,"_156":141,"_227":300},16,["D",1747923387722],[301],{"_107":302,"_156":141,"_154":185,"_100":184,"_164":303,"_162":178,"_104":290,"_155":-5,"_109":299,"_157":298,"_167":-5,"_158":27,"_166":27,"_103":304},"59b810e8-3717-11f0-afd5-9e149126039e","import agent from \"./agent.tsx\"\n\nexport default async function(req: Request) {\n // Default objective as fallback\n let objective = `Buy 100 shares of MSFT and 100 shares of NVDA.`;\n\n try {\n if (req.method === \"POST\") {\n // Check if the request is a form submission or JSON\n const contentType = req.headers.get(\"Content-Type\") || \"\";\n\n if (contentType.includes(\"application/json\")) {\n const body = await req.json();\n return new Response(JSON.stringify(body, null, 2), {\n headers: {\n \"Content-Type\": \"application/json\",\n },\n });\n } else if (contentType.includes(\"application/x-www-form-urlencoded\")) {\n // Process form data\n const formData = await req.formData();\n objective = formData.get(\"objective\")?.toString() || objective;\n\n // Continue with API call using the submitted objective\n } else {\n return new Response(\"Unsupported content type\", { status: 415 });\n }\n\n // Make API call with the objective from the form\n const requestBody = {\n messages: [\n { role: \"user\", content: objective },\n ],\n streamResults: false,\n };\n\n const response = await agent(new Request(\"/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n });\n\n if (!response.ok) {\n throw new Error(`HTTP error! Status: ${response.status}, Text: ${await response.text()}`);\n }\n\n // Get the API response data\n const responseData = await response.json();\n console.log(\"API Response:\", responseData);\n\n // Return HTML with the results\n return new Response(\n generateHtmlPage(objective, responseData),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n } else {\n // For GET requests, return the HTML form\n return new Response(\n generateHtmlPage(),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n }\n } catch (error) {\n console.error(\"Error:\", error);\n // Return error as HTML\n return new Response(\n generateHtmlPage(objective, null, `Failed to fetch results: ${error.message}`),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n status: 500,\n },\n );\n }\n}\n\n// Function to generate HTML for tool calls\nfunction generateToolCallsHtml(results: any): string {\n if (!results || !results.steps) {\n return \"\";\n }\n\n // Collect all tool calls from all steps\n const allToolCalls: any[] = [];\n\n results.steps.forEach((step: any) => {\n if (step.toolResults && step.toolResults.length > 0) {\n step.toolResults.forEach((toolResult: any) => {\n allToolCalls.push(toolResult);\n });\n }\n });\n\n if (allToolCalls.length === 0) {\n return \"\";\n }\n\n // Generate HTML for each tool call\n const toolCallsHtml = allToolCalls.map((toolCall: any, index: number) => {\n const toolName = toolCall.name || \"Unknown Tool\";\n const args = toolCall.args || {};\n const result = toolCall.result || {};\n\n return `\n

\n ${toolName}\n

Arguments:

${JSON.stringify(args, null, 2)}

\n \n

Result:

${JSON.stringify(result, null, 2)}

\n `;\n }).join(\"\");\n\n return `\n

Tool Calls:

\n ${toolCallsHtml}\n

Autonomous Val

\n \n \n \n ${\n errorMessage\n ? `\n

Error

${errorMessage}

\n `\n : \"\"\n }\n \n ${\n results\n ? `\n

Results:

\n \n

${results.text || \"No text response\"}

\n \n ${generateToolCallsHtml(results)}\n

\n `\n : \"\"\n }\n

\n\n`;\n}",[184],{"_154":119,"_104":306,"_109":307,"_156":141,"_227":308},15,["D",1747920165913],[309],{"_107":310,"_156":141,"_154":119,"_100":153,"_164":311,"_162":163,"_104":213,"_155":-5,"_109":307,"_157":306,"_167":-5,"_158":27,"_166":27,"_103":312},"d95c251c-370f-11f0-9cd7-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n![image](https://private-user-images.githubusercontent.com/207076/446343060-67bf14c0-7ad4-4bb7-a98b-eee3eed48e31.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3NDc4ODIwMTEsIm5iZiI6MTc0Nzg4MTcxMSwicGF0aCI6Ii8yMDcwNzYvNDQ2MzQzMDYwLTY3YmYxNGMwLTdhZDQtNGJiNy1hOThiLWVlZTNlZWQ0OGUzMS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjUwNTIyJTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI1MDUyMlQwMjQxNTFaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT02ZTIwOTlmMTJhNDY2ZTAzMTJjNmUxMjA4ZGUyMjljYTU3NzZjMDA5MTFmNmJjMGY0ZGY5YTE2OGJhZWYyNjBlJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.3NR5pU8HyClzuL0dxamQ6YHF65WWYE33suS9-jaWem4)\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK, though not all models support tool calling. \n\nAnthropic's models should work well, but the system prompt would need to be retuned to work with Claude.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models\n\n## Plug\nAre you a Senior Software Engineer who wants to work on AI or TypeScript? Feel free to contact me to learn more about careers at AngelList. We're solving some of the hardest problems in Venture Capital and Private Markets and are always looking for great people to join us.",[153],{"_154":119,"_104":314,"_109":315,"_156":141,"_227":316},14,["D",1747882283753],[317],{"_107":318,"_156":141,"_154":119,"_100":153,"_164":319,"_162":163,"_104":203,"_155":-5,"_109":315,"_157":314,"_167":-5,"_158":27,"_166":27,"_103":320},"a5d4e3da-36b7-11f0-995b-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n![image](https://private-user-images.githubusercontent.com/207076/446343060-67bf14c0-7ad4-4bb7-a98b-eee3eed48e31.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3NDc4ODIwMTEsIm5iZiI6MTc0Nzg4MTcxMSwicGF0aCI6Ii8yMDcwNzYvNDQ2MzQzMDYwLTY3YmYxNGMwLTdhZDQtNGJiNy1hOThiLWVlZTNlZWQ0OGUzMS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjUwNTIyJTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI1MDUyMlQwMjQxNTFaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT02ZTIwOTlmMTJhNDY2ZTAzMTJjNmUxMjA4ZGUyMjljYTU3NzZjMDA5MTFmNmJjMGY0ZGY5YTE2OGJhZWYyNjBlJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.3NR5pU8HyClzuL0dxamQ6YHF65WWYE33suS9-jaWem4)\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models\n\n## Plug\nAre you a Senior Software Engineer who wants to work on AI or TypeScript? Feel free to contact me to learn more about careers at AngelList. We're solving some of the hardest problems in Venture Capital and Private Markets and are always looking for great people to join us.",[153],{"_154":185,"_104":322,"_109":323,"_156":141,"_227":324},13,["D",1747881852288],[325],{"_107":326,"_156":141,"_154":185,"_100":184,"_164":327,"_162":178,"_104":213,"_155":-5,"_109":323,"_157":322,"_167":-5,"_158":27,"_166":27,"_103":328},"a4a895ca-36b6-11f0-b75a-9e149126039e","export default async function(req: Request) {\n // Default objective as fallback\n let objective = `Buy 100 shares of MSFT and 100 shares of NVDA.`;\n\n try {\n if (req.method === \"POST\") {\n // Check if the request is a form submission or JSON\n const contentType = req.headers.get(\"Content-Type\") || \"\";\n\n if (contentType.includes(\"application/json\")) {\n const body = await req.json();\n return new Response(JSON.stringify(body, null, 2), {\n headers: {\n \"Content-Type\": \"application/json\",\n },\n });\n } else if (contentType.includes(\"application/x-www-form-urlencoded\")) {\n // Process form data\n const formData = await req.formData();\n objective = formData.get(\"objective\")?.toString() || objective;\n\n // Continue with API call using the submitted objective\n } else {\n return new Response(\"Unsupported content type\", { status: 415 });\n }\n\n // Make API call with the objective from the form\n const requestBody = {\n messages: [\n { role: \"user\", content: objective },\n ],\n streamResults: false,\n };\n\n const response = await fetch(\"https://abrinz--3be1cc2632ad11f080f5569c3dd06744.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n });\n\n if (!response.ok) {\n throw new Error(`HTTP error! Status: ${response.status}, Text: ${await response.text()}`);\n }\n\n // Get the API response data\n const responseData = await response.json();\n console.log(\"API Response:\", responseData);\n\n // Return HTML with the results\n return new Response(\n generateHtmlPage(objective, responseData),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n } else {\n // For GET requests, return the HTML form\n return new Response(\n generateHtmlPage(),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n }\n } catch (error) {\n console.error(\"Error:\", error);\n // Return error as HTML\n return new Response(\n generateHtmlPage(objective, null, `Failed to fetch results: ${error.message}`),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n status: 500,\n },\n );\n }\n}\n\n// Function to generate HTML for tool calls\nfunction generateToolCallsHtml(results: any): string {\n if (!results || !results.steps) {\n return \"\";\n }\n\n // Collect all tool calls from all steps\n const allToolCalls: any[] = [];\n\n results.steps.forEach((step: any) => {\n if (step.toolResults && step.toolResults.length > 0) {\n step.toolResults.forEach((toolResult: any) => {\n allToolCalls.push(toolResult);\n });\n }\n });\n\n if (allToolCalls.length === 0) {\n return \"\";\n }\n\n // Generate HTML for each tool call\n const toolCallsHtml = allToolCalls.map((toolCall: any, index: number) => {\n const toolName = toolCall.name || \"Unknown Tool\";\n const args = toolCall.args || {};\n const result = toolCall.result || {};\n\n return `\n

\n ${toolName}\n

Arguments:

${JSON.stringify(args, null, 2)}

\n \n

Result:

${JSON.stringify(result, null, 2)}

\n `;\n }).join(\"\");\n\n return `\n

Tool Calls:

\n ${toolCallsHtml}\n

Autonomous Val

\n \n \n \n ${\n errorMessage\n ? `\n

Error

${errorMessage}

\n `\n : \"\"\n }\n \n ${\n results\n ? `\n

Results:

\n \n

${results.text || \"No text response\"}

\n \n ${generateToolCallsHtml(results)}\n

\n `\n : \"\"\n }\n

\n\n`;\n}",[184],{"_154":185,"_104":330,"_109":331,"_156":141,"_227":332},12,["D",1747881829034],[333],{"_107":334,"_156":141,"_154":185,"_100":184,"_164":335,"_162":178,"_104":203,"_155":-5,"_109":331,"_157":330,"_167":-5,"_158":27,"_166":27,"_103":336},"96cc5162-36b6-11f0-98fc-9e149126039e","export default async function(req: Request) {\n // Default objective as fallback\n let objective = `Buy 100 shares of MSFT and 100 shares of NVDA.`;\n\n try {\n if (req.method === \"POST\") {\n // Check if the request is a form submission or JSON\n const contentType = req.headers.get(\"Content-Type\") || \"\";\n\n if (contentType.includes(\"application/json\")) {\n const body = await req.json();\n return new Response(JSON.stringify(body, null, 2), {\n headers: {\n \"Content-Type\": \"application/json\",\n },\n });\n } else if (contentType.includes(\"application/x-www-form-urlencoded\")) {\n // Process form data\n const formData = await req.formData();\n objective = formData.get(\"objective\")?.toString() || objective;\n\n // Continue with API call using the submitted objective\n } else {\n return new Response(\"Unsupported content type\", { status: 415 });\n }\n\n // Make API call with the objective from the form\n const requestBody = {\n messages: [\n { role: \"user\", content: objective },\n ],\n streamResults: false,\n };\n\n const response = await fetch(\"https://abrinz--3be1cc2632ad11f080f5569c3dd06744.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n });\n\n if (!response.ok) {\n throw new Error(`HTTP error! Status: ${response.status}, Text: ${await response.text()}`);\n }\n\n // Get the API response data\n const responseData = await response.json();\n console.log(\"API Response:\", responseData);\n\n // Return HTML with the results\n return new Response(\n generateHtmlPage(objective, responseData),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n } else {\n // For GET requests, return the HTML form\n return new Response(\n generateHtmlPage(),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n }\n } catch (error) {\n console.error(\"Error:\", error);\n // Return error as HTML\n return new Response(\n generateHtmlPage(objective, null, `Failed to fetch results: ${error.message}`),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n status: 500,\n },\n );\n }\n}\n\n// Function to generate HTML for tool calls\nfunction generateToolCallsHtml(results: any): string {\n if (!results || !results.steps) {\n return \"\";\n }\n\n // Collect all tool calls from all steps\n const allToolCalls: any[] = [];\n\n results.steps.forEach((step: any) => {\n if (step.toolResults && step.toolResults.length > 0) {\n step.toolResults.forEach((toolResult: any) => {\n allToolCalls.push(toolResult);\n });\n }\n });\n\n if (allToolCalls.length === 0) {\n return \"\";\n }\n\n // Generate HTML for each tool call\n const toolCallsHtml = allToolCalls.map((toolCall: any, index: number) => {\n const toolName = toolCall.name || \"Unknown Tool\";\n const args = toolCall.args || {};\n const result = toolCall.result || {};\n\n return `\n

\n ${toolName}\n

Arguments:

${JSON.stringify(args, null, 2)}

\n \n

Result:

${JSON.stringify(result, null, 2)}

\n `;\n }).join(\"\");\n\n return `\n

Tool Calls:

\n ${toolCallsHtml}\n

Autonomous Val Demo

\n \n \n \n ${\n errorMessage\n ? `\n

Error

${errorMessage}

\n `\n : \"\"\n }\n \n ${\n results\n ? `\n

Results:

\n \n

${results.text || \"No text response\"}

\n \n ${generateToolCallsHtml(results)}\n

\n `\n : \"\"\n }\n

\n\n`;\n}",[184],{"_154":119,"_104":338,"_109":339,"_156":141,"_227":340},11,["D",1747881787094],[341],{"_107":342,"_156":141,"_154":119,"_100":153,"_164":343,"_162":163,"_104":177,"_155":-5,"_109":339,"_157":338,"_167":-5,"_158":27,"_166":27,"_103":344},"7dccdc90-36b6-11f0-a5fc-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n![image](https://private-user-images.githubusercontent.com/207076/446343060-67bf14c0-7ad4-4bb7-a98b-eee3eed48e31.png?jwt=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpc3MiOiJnaXRodWIuY29tIiwiYXVkIjoicmF3LmdpdGh1YnVzZXJjb250ZW50LmNvbSIsImtleSI6ImtleTUiLCJleHAiOjE3NDc4ODIwMTEsIm5iZiI6MTc0Nzg4MTcxMSwicGF0aCI6Ii8yMDcwNzYvNDQ2MzQzMDYwLTY3YmYxNGMwLTdhZDQtNGJiNy1hOThiLWVlZTNlZWQ0OGUzMS5wbmc_WC1BbXotQWxnb3JpdGhtPUFXUzQtSE1BQy1TSEEyNTYmWC1BbXotQ3JlZGVudGlhbD1BS0lBVkNPRFlMU0E1M1BRSzRaQSUyRjIwMjUwNTIyJTJGdXMtZWFzdC0xJTJGczMlMkZhd3M0X3JlcXVlc3QmWC1BbXotRGF0ZT0yMDI1MDUyMlQwMjQxNTFaJlgtQW16LUV4cGlyZXM9MzAwJlgtQW16LVNpZ25hdHVyZT02ZTIwOTlmMTJhNDY2ZTAzMTJjNmUxMjA4ZGUyMjljYTU3NzZjMDA5MTFmNmJjMGY0ZGY5YTE2OGJhZWYyNjBlJlgtQW16LVNpZ25lZEhlYWRlcnM9aG9zdCJ9.3NR5pU8HyClzuL0dxamQ6YHF65WWYE33suS9-jaWem4)\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models\n\n## Plug\nAre you a Senior Software Engineer who wants to work on AI or TypeScript? Feel free to DM me on the Val Town Discord to learn more about careers at AngelList. We're solving some of the hardest problems in Venture Capital and Private Markets and are always looking for great people to join us.",[153],{"_154":185,"_104":346,"_109":347,"_156":141,"_227":348},10,["D",1747863611421],[349],{"_107":350,"_156":141,"_154":185,"_100":184,"_164":351,"_162":178,"_104":177,"_155":-5,"_109":347,"_157":346,"_167":-5,"_158":27,"_166":27,"_103":352},"2c491e6c-368c-11f0-bbbb-9e149126039e","export default async function(req: Request) {\n // Default objective as fallback\n let objective = `Buy 100 shares of MSFT and 100 shares of NVDA.`;\n\n try {\n if (req.method === \"POST\") {\n // Check if the request is a form submission or JSON\n const contentType = req.headers.get(\"Content-Type\") || \"\";\n\n if (contentType.includes(\"application/json\")) {\n const body = await req.json();\n return new Response(JSON.stringify(body, null, 2), {\n headers: {\n \"Content-Type\": \"application/json\",\n },\n });\n } else if (contentType.includes(\"application/x-www-form-urlencoded\")) {\n // Process form data\n const formData = await req.formData();\n objective = formData.get(\"objective\")?.toString() || objective;\n\n // Continue with API call using the submitted objective\n } else {\n return new Response(\"Unsupported content type\", { status: 415 });\n }\n\n // Make API call with the objective from the form\n const requestBody = {\n messages: [\n { role: \"user\", content: objective },\n ],\n streamResults: false,\n };\n\n const response = await fetch(\"https://abrinz--3be1cc2632ad11f080f5569c3dd06744.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n });\n\n if (!response.ok) {\n throw new Error(`HTTP error! Status: ${response.status}, Text: ${await response.text()}`);\n }\n\n // Get the API response data\n const responseData = await response.json();\n console.log(\"API Response:\", responseData);\n\n // Return HTML with the results\n return new Response(\n generateHtmlPage(objective, responseData),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n } else {\n // For GET requests, return the HTML form\n return new Response(\n generateHtmlPage(),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n },\n );\n }\n } catch (error) {\n console.error(\"Error:\", error);\n // Return error as HTML\n return new Response(\n generateHtmlPage(objective, null, `Failed to fetch results: ${error.message}`),\n {\n headers: {\n \"Content-Type\": \"text/html\",\n },\n status: 500,\n },\n );\n }\n}\n\n// Function to generate HTML for tool calls\nfunction generateToolCallsHtml(results: any): string {\n if (!results || !results.steps) {\n return \"\";\n }\n\n // Collect all tool calls from all steps\n const allToolCalls: any[] = [];\n\n results.steps.forEach((step: any) => {\n if (step.toolResults && step.toolResults.length > 0) {\n step.toolResults.forEach((toolResult: any) => {\n allToolCalls.push(toolResult);\n });\n }\n });\n\n if (allToolCalls.length === 0) {\n return \"\";\n }\n\n // Generate HTML for each tool call\n const toolCallsHtml = allToolCalls.map((toolCall: any, index: number) => {\n const toolName = toolCall.name || \"Unknown Tool\";\n const args = toolCall.args || {};\n const result = toolCall.result || {};\n\n return `\n

\n ${toolName}\n

Arguments:

${JSON.stringify(args, null, 2)}

\n \n

Result:

${JSON.stringify(result, null, 2)}

\n `;\n }).join(\"\");\n\n return `\n

Tool Calls:

\n ${toolCallsHtml}\n

Val Town Hall Agent Demo

\n \n \n \n ${\n errorMessage\n ? `\n

Error

${errorMessage}

\n `\n : \"\"\n }\n \n ${\n results\n ? `\n

Results:

\n \n

${results.text || \"No text response\"}

\n \n ${generateToolCallsHtml(results)}\n

\n `\n : \"\"\n }\n

\n\n`;\n}",[184],{"_154":193,"_104":194,"_109":354,"_156":141,"_227":355},["D",1747863611398],[356],{"_107":191,"_156":141,"_154":193,"_100":192,"_164":196,"_162":178,"_104":177,"_155":-5,"_109":354,"_157":194,"_167":-5,"_158":159,"_166":27,"_103":357},[192],{"_154":174,"_104":175,"_109":359,"_156":141,"_227":360},["D",1747863611397],[361],{"_107":172,"_156":141,"_154":174,"_100":173,"_164":179,"_162":178,"_104":177,"_155":-5,"_109":359,"_157":175,"_167":-5,"_158":159,"_166":27,"_103":362},[173],{"_154":221,"_104":222,"_109":364,"_156":141,"_227":365},["D",1747863611211],[366],{"_107":219,"_156":141,"_154":221,"_100":220,"_164":224,"_162":205,"_104":31,"_155":-5,"_109":364,"_157":222,"_167":-5,"_158":159,"_166":27,"_103":367},[220],{"_154":185,"_104":161,"_109":369,"_156":141,"_227":370},["D",1747863611040],[371],{"_107":372,"_156":141,"_154":185,"_100":184,"_164":351,"_162":205,"_104":31,"_155":-5,"_109":369,"_157":161,"_167":-5,"_158":27,"_166":27,"_103":373},"2c1f136a-368c-11f0-a367-9e149126039e",[184],{"_154":193,"_104":124,"_109":375,"_156":141,"_227":376},["D",1747863611176],[377],{"_107":378,"_156":141,"_154":193,"_100":192,"_164":196,"_162":205,"_104":31,"_155":-5,"_109":375,"_157":124,"_167":-5,"_158":27,"_166":27,"_103":379},"2c1b51e4-368c-11f0-911e-9e149126039e",[192],{"_154":119,"_104":290,"_109":381,"_156":141,"_227":382},["D",1747863611098],[383],{"_107":384,"_156":141,"_154":119,"_100":153,"_164":385,"_162":163,"_104":31,"_155":-5,"_109":381,"_157":290,"_167":-5,"_158":27,"_166":27,"_103":386},"2c111760-368c-11f0-8f1a-9e149126039e","# Autonomous Val\nThis project demonstrates how to build autonomous agents on Val Town that can be triggered by API calls, cron jobs, etc.\n\n## Setup\nRemix this Val into your Val Town account.\n\nConfigure the following variables in your environment: \n- `AGENT_API_KEY` (This is a secure token that you choose to secure the agent.tsx POST endpoint)\n- `OPENAI_API_KEY` (An OpenAI API Key)\n- `EXA_API_KEY` (Optional, though needed if you use the web search tool)\n\n## Usage\nUse `demo.tsx` to send objectives to your agent. \n\n### API Usage\nTo use the API from another client, you can POST authenticated requests to the agent.tsx endpoint:\n\n```javascript\nconst requestBody = {\n messages: [\n { role: \"user\", content: \"What is the difference in height between the Salesforce Tower and the Empire State Building?\" },\n ],\n streamResults: false,\n};\n\nconst response = await fetch(\"https://.web.val.run/\", {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"Authorization\": `Bearer ${Deno.env.get(\"AGENT_API_KEY\")}`,\n },\n body: JSON.stringify(requestBody),\n});\n```\n\n### Streaming Chat\nThe API will also work with streaming chat front ends based on the Vercel AI SDK's useChat hook.\n\nYou just need to pass `streamResults: true` in your API POST request.\n\n## Using Other Models\nIt is easy to reconfigure the agent to work with other providers that are supported by Vercel's AI SDK.\n\nLearn more: https://ai-sdk.dev/docs/foundations/providers-and-models\n\n## Plug\nAre you a Senior Software Engineer who wants to work on AI or TypeScript? Feel free to DM me on the Val Town Discord to learn more about careers at AngelList. We're solving some of the hardest problems in Venture Capital and Private Markets and are always looking for great people to join us.",[153],{"_154":212,"_104":213,"_109":388,"_156":141,"_227":389},["D",1747863611098],[390],{"_107":210,"_156":141,"_154":212,"_100":211,"_164":215,"_162":205,"_104":31,"_155":-5,"_109":388,"_157":213,"_167":-5,"_158":159,"_166":27,"_103":391},[211],{"_154":202,"_104":203,"_109":393,"_156":141,"_227":394},["D",1747863611040],[395],{"_107":200,"_156":141,"_154":202,"_100":201,"_164":206,"_162":205,"_104":31,"_155":-5,"_109":393,"_157":203,"_167":-5,"_158":159,"_166":27,"_103":396},[201],{"_154":174,"_104":177,"_109":398,"_156":141,"_227":399},["D",1747863611040],[400],{"_107":401,"_156":141,"_154":174,"_100":173,"_164":179,"_162":205,"_104":31,"_155":-5,"_109":398,"_157":177,"_167":-5,"_158":27,"_166":27,"_103":402},"2c0729d0-368c-11f0-b862-9e149126039e",[173],"page","hasNext","routes/_app","actionData","errors"]

History