Autonomous Agent Framework

Goal Inbox

\n \n

\n\n

Agent Control

💤

\n Agent Power\n

\n Off / On\n \n \n \n \n

\n Autopilot\n

\n Off / On\n \n \n \n \n

\n Autopilot Mission\n \n

Awaiting approval for the proposed plan.

\n \n \n

\n\n

Application State

\n\n

Forensic Debug Log

\n\n

Data Management

\n Clear specific categories of application data.\n

\n \n \n \n \n

\n\n\n\n\n\n`;\n}\n\n// --- BACKEND API HANDLER ---\nexport default async function (req: Request) {\n const url = new URL(req.url);\n const action = url.searchParams.get(\"action\");\n\n if (req.method === \"GET\") {\n return new Response(generateHtml(req.url), {\n headers: { \"Content-Type\": \"text/html\" },\n });\n }\n\n if (req.method === \"POST\") {\n try {\n const openai = new OpenAI();\n const body = await req.json();\n\n switch (action) {\n case \"decide\": {\n const completion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: SYSTEM_PROMPT },\n { role: \"user\", content: body.context },\n ],\n tools: body.tools.map((t) => ({ type: \"function\", function: t })),\n tool_choice: \"auto\",\n response_format: { type: \"json_object\" },\n });\n const responseMessage = completion.choices[0].message;\n const toolCall = responseMessage.tool_calls[0].function;\n return new Response(\n JSON.stringify({\n tool: toolCall.name,\n args: JSON.parse(toolCall.arguments),\n }),\n { headers: { \"Content-Type\": \"application/json\" } },\n );\n }\n\n case \"cureGoal\": {\n const completion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: GOAL_CURING_PROMPT },\n {\n role: \"user\",\n content: `User Goal: \"${body.goal}\"\\nAvailable Tools: ${\n JSON.stringify(body.tools)\n }`,\n },\n ],\n response_format: { type: \"json_object\" },\n });\n return new Response(completion.choices[0].message.content, {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n case \"reflect\": {\n const completion = await openai.chat.completions.create({\n model: \"gpt-4o\",\n messages: [\n { role: \"system\", content: SELF_CORRECTION_PROMPT },\n {\n role: \"user\",\n content: `Original Goal: ${\n JSON.stringify(body.goal)\n }\\nFailed Plan: ${\n JSON.stringify(body.failedPlan)\n }\\nError Message: \"${body.error}\"\\nAvailable Tools: ${\n JSON.stringify(body.tools)\n }`,\n },\n ],\n response_format: { type: \"json_object\" },\n });\n return new Response(completion.choices[0].message.content, {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n // --- Long-Term Memory Simulation using std/blob ---\n // NOTE: This is a simplified simulation. A production system would use a proper vector database.\n case \"addExperience\": {\n const experiences = await blob.getJSON(\"ltm_experiences_v2\") || [];\n experiences.push({\n timestamp: new Date().toISOString(),\n text: body.text,\n });\n if (experiences.length > 50) experiences.shift(); // Cap memory size\n await blob.setJSON(\"ltm_experiences_v2\", experiences);\n return new Response(JSON.stringify({ success: true }), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n case \"queryExperiences\": {\n const experiences = await blob.getJSON(\"ltm_experiences_v2\") || [];\n // Simple retrieval of the most recent experiences. A real LTM would perform semantic search.\n const relevant = experiences.slice(-body.count);\n return new Response(JSON.stringify(relevant), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n case \"clearExperiences\": {\n await blob.delete(\"ltm_experiences_v2\");\n return new Response(JSON.stringify({ success: true }), {\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n\n default:\n return new Response(JSON.stringify({ error: \"Invalid action\" }), {\n status: 400,\n });\n }\n } catch (error) {\n console.error(`API Error on action '${action}':`, error);\n return new Response(JSON.stringify({ error: error.message }), {\n status: 500,\n headers: { \"Content-Type\": \"application/json\" },\n });\n }\n }\n\n return new Response(\"Method Not Allowed\", { status: 405 });\n}","deleted","label",[98],"permalink","https://www.val.town/x/join/stocks/code/main.tsx","extension","actionData","errors"]

Goal Inbox

Agent Control

Application State

Forensic Debug Log

Data Management

Confirm Action