1# Research Agent over email
2
3This is a lightweight wrapper around Perplexity's web search API that you can interact with via email.
16 );
17
18 const client = new OpenAI({ apiKey: process.env.PERPLEXITY_API_KEY, baseURL: "https://api.perplexity.ai" });
19 const response = await client.chat.completions.create({
20 model: "sonar",
125 onChange={handlePathChange}
126 className="px-2 py-1 text-sm w-32 md:w-40"
127 placeholder="Path (e.g., /api/data)"
128 />
129 <button
8import { ChatMessages } from "./ChatMessages.tsx";
9import { ChatInput } from "./ChatInput.tsx";
10import { ApiKeyWarning } from "./ApiKeyWarning.tsx";
11import { processFiles } from "./ImageUpload.tsx";
12import { Preview } from "./Preview.tsx";
15 project,
16 bearerToken,
17 anthropicApiKey,
18 setProject,
19}: {
20 project: any;
21 bearerToken: string;
22 anthropicApiKey: string;
23 setProject: (project: any) => void;
24}) {
57 project,
58 branchId,
59 anthropicApiKey,
60 bearerToken,
61 selectedFiles,
170
171 <div className="p-6 flex flex-col h-full w-full">
172 <ApiKeyWarning show={!anthropicApiKey} />
173
174 <div className="flex flex-col lg:flex-row gap-4">
72. Send notifications to your preferred platform (Discord, Slack, email, etc.)
8
9Reddit does not have an API that allows users to scrape data, so we are doing this with the Google Search API, [Serp](https://serpapi.com/).
10
11---
29
30---
31### 3. Get a SerpApi Key
32This template requires a [SerpApi](https://serpapi.com/) key to search Reddit posts via Google search results.
33
341. **Get a SerpApi key**:
35 - Sign up at [SerpApi](https://serpapi.com/) to create an account.
36 - Generate an API key from your account dashboard.
37
382. **Add the SerpApi key to your environment variables**:
39 - Go to your [Val Town environment variables](https://www.val.town/settings/environment-variables).
40 - Add a new key:
41 - Key: `SERP_API_KEY`
42 - Value: Your SERP API key.
43
44Without this key, the val will not function correctly.
75
76### NOTE: Usage Limits
77- **SerpApi:** Free SerpApi accounts have monthly call limits.
97
98export async function sendDailyBriefing(chatId?: string, today?: DateTime) {
99 // Get API keys from environment
100 const apiKey = Deno.env.get("ANTHROPIC_API_KEY");
101 const telegramToken = Deno.env.get("TELEGRAM_TOKEN");
102
106 }
107
108 if (!apiKey) {
109 console.error("Anthropic API key is not configured.");
110 return;
111 }
122
123 // Initialize Anthropic client
124 const anthropic = new Anthropic({ apiKey });
125
126 // Initialize Telegram bot
162
163 // disabled title for now, it seemes unnecessary...
164 // await bot.api.sendMessage(chatId, `*${title}*`, { parse_mode: "Markdown" });
165
166 // Then send the main content
169
170 if (content.length <= MAX_LENGTH) {
171 await bot.api.sendMessage(chatId, content, { parse_mode: "Markdown" });
172 // Store the briefing in chat history
173 await storeChatMessage(
198 // Send each chunk as a separate message and store in chat history
199 for (const chunk of chunks) {
200 await bot.api.sendMessage(chatId, chunk, { parse_mode: "Markdown" });
201 // Store each chunk in chat history
202 await storeChatMessage(
53You'll need to set up some environment variables to make it run.
54
55- `ANTHROPIC_API_KEY` for LLM calls
56- You'll need to follow [these instructions](https://docs.val.town/integrations/telegram/) to make a telegram bot, and set `TELEGRAM_TOKEN`. You'll also need to get a `TELEGRAM_CHAT_ID` in order to have the bot remember chat contents.
57- For the Google Calendar integration you'll need `GOOGLE_CALENDAR_ACCOUNT_ID` and `GOOGLE_CALENDAR_CALENDAR_ID`. See [these instuctions](https://www.val.town/v/stevekrouse/pipedream) for details.
8## Hono
9
10This app uses [Hono](https://hono.dev/) as the API framework. You can think of Hono as a replacement for [ExpressJS](https://expressjs.com/) that works in serverless environments like Val Town or Cloudflare Workers. If you come from Python or Ruby, Hono is also a lot like [Flask](https://github.com/pallets/flask) or [Sinatra](https://github.com/sinatra/sinatra), respectively.
11
12## Serving assets to the frontend
20### `index.html`
21
22The most complicated part of this backend API is serving index.html. In this app (like most apps) we serve it at the root, ie `GET /`.
23
24We *bootstrap* `index.html` with some initial data from the server, so that it gets dynamically injected JSON data without having to make another round-trip request to the server to get that data on the frontend. This is a common pattern for client-side rendered apps.
25
26## CRUD API Routes
27
28This app has two CRUD API routes: for reading and inserting into the messages table. They both speak JSON, which is standard. They import their functions from `/backend/database/queries.ts`. These routes are called from the React app to refresh and update data.
29
30## Errors
31
32Hono and other API frameworks have a habit of swallowing up Errors. We turn off this default behavior by re-throwing errors, because we think most of the time you'll want to see the full stack trace instead of merely "Internal Server Error". You can customize how you want errors to appear.
8import { type Memory } from "../../shared/types.ts";
9
10const API_BASE = "/api/memories";
11const MEMORIES_PER_PAGE = 20;
12
71 setError(null);
72 try {
73 const response = await fetch(API_BASE);
74 if (!response.ok) {
75 throw new Error(`HTTP error! status: ${response.status}`);
100
101 try {
102 const response = await fetch(API_BASE, {
103 method: "POST",
104 headers: { "Content-Type": "application/json" },
123
124 try {
125 const response = await fetch(`${API_BASE}/${id}`, {
126 method: "DELETE",
127 });
155
156 try {
157 const response = await fetch(`${API_BASE}/${editingMemory.id}`, {
158 method: "PUT",
159 headers: { "Content-Type": "application/json" },
18});
19
20// --- API Routes for Memories ---
21
22// GET /api/memories - Retrieve all memories
23app.get("/api/memories", async (c) => {
24 const memories = await getAllMemories();
25 return c.json(memories);
26});
27
28// POST /api/memories - Create a new memory
29app.post("/api/memories", async (c) => {
30 const body = await c.req.json<Omit<Memory, "id">>();
31 if (!body.text) {
36});
37
38// PUT /api/memories/:id - Update an existing memory
39app.put("/api/memories/:id", async (c) => {
40 const id = c.req.param("id");
41 const body = await c.req.json<Partial<Omit<Memory, "id">>>();
58});
59
60// DELETE /api/memories/:id - Delete a memory
61app.delete("/api/memories/:id", async (c) => {
62 const id = c.req.param("id");
63 try {
75// --- Blob Image Serving Routes ---
76
77// GET /api/images/:filename - Serve images from blob storage
78app.get("/api/images/:filename", async (c) => {
79 const filename = c.req.param("filename");
80
