AI Slackbot

BYOA Slack agent. Click below to create the Slack app — scopes, events, and webhook URL are pre-filled from a manifest.

Create Slack App →

Then add SLACK_BOT_TOKEN, SLACK_SIGNING_SECRET, and ANTHROPIC_API_KEY to your val's environment variables. See the README.

\n`);\n});\n\n// ─── Slack event intake: verify, then dispatch to /process ──────────────────\n// Val Town kills detached promises after the HTTP response is sent, so the\n// fire-and-forget pattern truncates slow handlers (long threads + tool calls).\n// We use the \"early return\" pattern: dispatch heavy work to /process as a\n// separate HTTP request with its own full execution lifecycle.\n// See: https://docs.val.town/vals/http/early-return/\n\napp.post(\"/events\", async (c) => {\n const rawBody = await c.req.text();\n const payload = JSON.parse(rawBody);\n const requestType = payload.type as \"url_verification\" | \"event_callback\";\n\n // See https://api.slack.com/events/url_verification\n if (requestType === \"url_verification\") {\n return new Response(payload.challenge, { status: 200 });\n }\n\n try {\n await verifyRequest({ requestType, request: c.req.raw, rawBody });\n const botUserId = await getBotId();\n const event = payload.event as SlackEvent;\n console.log(`[Events] Received ${event.type}, dispatching to /process`);\n\n // Fire off a NEW HTTP request to /process — this gets its own execution\n // lifecycle. Include an internal secret so /process rejects unauthorized\n // callers.\n const processUrl = new URL(\"/process\", c.req.url).toString();\n const internalSecret = Deno.env.get(\"SLACK_SIGNING_SECRET\")!;\n fetch(processUrl, {\n method: \"POST\",\n headers: {\n \"Content-Type\": \"application/json\",\n \"X-Internal-Secret\": internalSecret,\n },\n body: JSON.stringify({ event, botUserId }),\n }); // intentionally NOT awaited\n\n return new Response(\"OK\", { status: 200 });\n } catch (error) {\n console.error(\"[Events] Error:\", error);\n return new Response(\"Error\", { status: 500 });\n }\n});\n\n// ─── Process route: does the actual heavy lifting ───────────────────────────\n// This runs as its own HTTP request, so it gets the full timeout window.\n\napp.post(\"/process\", async (c) => {\n // Verify this request came from our own /events handler\n const internalSecret = Deno.env.get(\"SLACK_SIGNING_SECRET\")!;\n const providedSecret = c.req.header(\"X-Internal-Secret\");\n if (providedSecret !== internalSecret) {\n console.warn(\"[Process] Rejected unauthorized request\");\n return new Response(\"Unauthorized\", { status: 401 });\n }\n\n const { event, botUserId } = await c.req.json();\n console.log(`[Process] Handling ${event.type}`);\n\n try {\n switch (event.type) {\n case \"app_mention\":\n await handleAppMention(event, botUserId);\n break;\n\n case \"assistant_thread_started\":\n await handleThreadStarted(event);\n break;\n\n case \"message\":\n await handleMessage(event as GenericMessageEvent, botUserId);\n break;\n\n default:\n console.log(`[Process] Unknown event type: ${event.type}`);\n }\n\n return new Response(\"Done\", { status: 200 });\n } catch (error) {\n console.error(`[Process] FAILED for ${event.type}:`, error);\n return new Response(\"Error\", { status: 500 });\n }\n});\n\nexport default app.fetch;\n",[125],"cb583f8e-481d-449b-abd2-50dd65339d02",["D",1782761585368],"# ai-slackbot\n\nAn AI agent powered by Claude with web search, deployed on Val Town.\n\n## Features\n\n- Responds to `@mentions` in channels and threads, with full thread history as\n context\n- Acts as a Slack assistant in DMs (Chat tab), with greeting + suggested prompts\n- Powered by Claude Sonnet 4.6 with Anthropic's built-in web search tool\n- One-click Slack app setup — no manual scope or event configuration\n- Easily extensible: swap models, add tools, change prompts\n\n## Setup (~2 minutes)\n\n1. Click [**Remix**](https://www.val.town/x/templates/ai-slackbot/?intent=remix)\n to get your own copy of this val.\n2. **Open your val's HTTP endpoint in a new tab** (you should see it in the\n homepage of your remixed val), then click **Create Slack App**. The app's\n [manifest](./events.ts) — scopes, events, and webhook URL — is pre-filled, so\n just click **Next** → pick workspace → **Create**.\n3. **Install to Workspace** in the Slack app's _Install App_ tab.\n4. **Paste 3 secrets** into your val's\n [environment variables](https://docs.val.town/reference/environment-variables/)\n (the slots are already created — just fill the values):\n - `SLACK_BOT_TOKEN` — Slack app → OAuth & Permissions → Bot User OAuth Token\n (`xoxb-...`)\n - `SLACK_SIGNING_SECRET` — Slack app → Basic Information → Signing Secret\n - `ANTHROPIC_API_KEY` — from the\n [Claude console](https://platform.claude.com/settings/keys)\n\n## Usage\n\n- **DM the bot** — open the Slack app, click _Messages_, and chat. The\n conversation maintains thread context automatically.\n- **`@mention` in a channel** — the bot replies in a thread on your message.\n Follow-ups in that thread also get full thread context.\n\n> ⚠️ **Invite the bot to the channel first** (`/invite @AI Slackbot`). Without\n> channel membership, the bot receives the mention event but can't read thread\n> history, and will only see your single mention as context.\n\n## Architecture\n\n```\nai-slackbot/\n├── events.ts # HTTP entry point: setup page + Slack event router\n├── prompt.md # System prompt for your agent\n├── README.md\n└── lib/\n ├── generate-response.ts # AI core (Claude + web search)\n ├── handle-app-mention.ts # @mentions in channels and threads\n ├── handle-message.ts # DMs in assistant (Chat tab) threads\n ├── handle-thread-started.ts # Greeting + suggested prompts on new assistant thread\n └── slack-utils.ts # Slack client, request verification, thread fetching\n```\n\nSlack POSTs events to `/events`. The handler verifies the signature, then\ndispatches the heavy work to `/process` as a separate HTTP request — Val Town\nkills detached promises after the response is sent, so the\n[early-return pattern](https://docs.val.town/vals/http/early-return/) gives the\nreal handler its own full execution lifecycle.\n\n## Customization\n\n- **Swap the model or model provider** → `lib/generate-response.ts` (uses the\n Vercel AI SDK, so any provider works)\n- **Edit the system prompt** → `prompt.md`\n- **Add tools** (database queries, knowledge bases, custom APIs) → extend the\n `tools` object in `lib/generate-response.ts`. See the\n [AI SDK tools docs](https://sdk.vercel.ai/docs/ai-sdk-core/tools-and-tool-calling).\n- **Change greeting or suggested prompts** in DMs →\n `lib/handle-thread-started.ts`\n- **Change Slack scopes, events, or app metadata** → edit the `manifest` object\n at the top of `events.ts`. Re-create the Slack app from the updated setup\n page.\n\n## Version pinning\n\nThis val uses\n[pinned imports](https://docs.val.town/reference/version-control/#imports--pinning)\nto avoid breaking changes:\n\n1. `@slack/web-api@7.14.1`\n2. `hono@4.12.0`\n3. `ai@6.0.92`\n4. `@ai-sdk/anthropic@3.0.45`\n\n## Credits\n\nBased on Vercel's\n[Slackbot Agent Guide](https://ai-sdk.dev/cookbook/guides/slackbot) and Pete\nMillspaugh's [pmillspaugh/duck](https://www.val.town/x/pmillspaugh/duck)",[119],"a7092403-f2b3-46f3-a0fb-c260e94d2fd8",["D",1782761585368],[112],"pulls","branches","forks","envVars",{"_127":170,"_249":170},"connections",{"_127":20,"_248":20},"unfilledCount","emptyCount",{"_27":238,"_13":112,"_74":111,"_37":-5,"_29":30,"_223":34,"_224":144,"_31":294,"_33":20,"_77":113,"_226":-5,"_79":24,"_228":-5,"_260":295,"_184":296},{"_27":289,"_13":76,"_74":75,"_37":111,"_29":30,"_223":34,"_224":144,"_31":290,"_33":20,"_77":78,"_226":291,"_79":24,"_228":-5,"_260":292,"_184":293},{"_27":284,"_13":88,"_74":87,"_37":111,"_29":30,"_223":34,"_224":144,"_31":285,"_33":20,"_77":78,"_226":286,"_79":24,"_228":-5,"_260":287,"_184":288},{"_27":279,"_13":94,"_74":93,"_37":111,"_29":30,"_223":34,"_224":144,"_31":280,"_33":20,"_77":78,"_226":281,"_79":24,"_228":-5,"_260":282,"_184":283},{"_27":274,"_13":100,"_74":99,"_37":111,"_29":30,"_223":34,"_224":144,"_31":275,"_33":20,"_77":78,"_226":276,"_79":24,"_228":-5,"_260":277,"_184":278},{"_27":269,"_13":106,"_74":105,"_37":111,"_29":30,"_223":34,"_224":144,"_31":270,"_33":20,"_77":78,"_226":271,"_79":24,"_228":-5,"_260":272,"_184":273},{"_27":234,"_13":119,"_74":118,"_37":-5,"_29":30,"_223":34,"_224":144,"_31":266,"_33":20,"_77":48,"_226":236,"_79":24,"_228":-5,"_260":267,"_184":268},{"_27":230,"_13":125,"_74":124,"_37":-5,"_29":30,"_223":34,"_224":144,"_31":263,"_33":20,"_77":126,"_226":232,"_79":24,"_228":-5,"_260":264,"_184":265},{"_27":222,"_13":131,"_74":130,"_37":-5,"_29":30,"_223":34,"_224":144,"_31":259,"_33":20,"_77":48,"_226":227,"_79":24,"_228":-5,"_260":261,"_184":262},["D",1782761585368],"children",[],[131],["D",1782761585368],[],[125],["D",1782761585368],[],[119],"0588bd4d-c001-4fa6-829c-416f9f37cb32",["D",1782761585368],"import { WebClient } from \"npm:@slack/web-api@7.14.1\";\nimport { ModelMessage } from \"npm:ai@6.0.92\";\n\nconst signingSecret = Deno.env.get(\"SLACK_SIGNING_SECRET\")!;\n\nexport const client = new WebClient(Deno.env.get(\"SLACK_BOT_TOKEN\"));\n\n// See https://api.slack.com/authentication/verifying-requests-from-slack\nexport async function isValidSlackRequest({\n request,\n rawBody,\n}: {\n request: Request;\n rawBody: string;\n}) {\n const timestamp = request.headers.get(\"X-Slack-Request-Timestamp\");\n const slackSignature = request.headers.get(\"X-Slack-Signature\");\n\n if (!timestamp || !slackSignature) {\n console.log(\"Missing timestamp or signature\");\n return false;\n }\n\n // Prevent replay attacks on the order of 5 minutes\n if (Math.abs(Date.now() / 1000 - parseInt(timestamp)) > 60 * 5) {\n console.log(\"Timestamp out of range\");\n return false;\n }\n\n const encoder = new TextEncoder();\n const base = `v0:${timestamp}:${rawBody}`;\n\n const key = await crypto.subtle.importKey(\n \"raw\",\n encoder.encode(signingSecret),\n { name: \"HMAC\", hash: \"SHA-256\" },\n false,\n [\"sign\"],\n );\n\n const signature = await crypto.subtle.sign(\n \"HMAC\",\n key,\n encoder.encode(base),\n );\n\n const hmacHex = Array.from(new Uint8Array(signature))\n .map((b) => b.toString(16).padStart(2, \"0\"))\n .join(\"\");\n const computedSignature = encoder.encode(`v0=${hmacHex}`);\n const expectedSignature = encoder.encode(slackSignature);\n\n // Prevent timing attacks with constant-time comparison\n if (computedSignature.length !== expectedSignature.length) return false;\n let result = 0;\n for (let i = 0; i < computedSignature.length; i++) {\n result |= computedSignature[i] ^ expectedSignature[i];\n }\n return result === 0;\n}\n\nexport const verifyRequest = async ({\n requestType,\n request,\n rawBody,\n}: {\n requestType: string;\n request: Request;\n rawBody: string;\n}) => {\n const validRequest = await isValidSlackRequest({ request, rawBody });\n if (!validRequest || requestType !== \"event_callback\") {\n throw new Error(\"Invalid request\");\n }\n};\n\nexport const updateStatus = async (\n { channel, thread_ts, status }: {\n channel: string;\n thread_ts: string;\n status: string;\n },\n) => {\n await client.assistant.threads.setStatus({\n channel_id: channel,\n thread_ts: thread_ts,\n status: status,\n loading_messages: [\n \"thinking...\",\n \"pondering...\",\n \"musing...\",\n \"ruminating...\",\n \"reflecting...\",\n \"contemplating...\",\n \"cogitating...\",\n \"mulling...\",\n \"noodling...\",\n \"deliberating...\",\n ],\n });\n};\n\nexport const updateThreadTitle = async (\n { channel, thread_ts, title }: {\n channel: string;\n thread_ts: string;\n title: string;\n },\n) => {\n await client.assistant.threads.setTitle({\n channel_id: channel,\n thread_ts: thread_ts,\n title: title,\n });\n};\n\nexport async function getThread(\n channel_id: string,\n thread_ts: string,\n botUserId: string,\n): Promise {\n const { messages } = await client.conversations.replies({\n channel: channel_id,\n ts: thread_ts,\n limit: 50,\n });\n\n if (!messages) throw new Error(\"No messages found in thread\");\n\n const result = messages\n .map((message) => {\n const isBot = !!message.bot_id;\n if (!message.text) return null;\n\n // For app mentions, remove the mention prefix\n // For IM messages, keep the full text\n let content = message.text;\n if (!isBot && content.includes(`<@${botUserId}>`)) {\n content = content.replace(`<@${botUserId}> `, \"\");\n }\n\n return {\n role: isBot ? \"assistant\" : \"user\",\n content: content,\n } as ModelMessage;\n })\n .filter((msg): msg is ModelMessage => msg !== null);\n\n return result;\n}\n\nexport const getBotId = async () => {\n const { user_id: botUserId } = await client.auth.test();\n\n if (!botUserId) {\n throw new Error(\"botUserId is undefined\");\n }\n return botUserId;\n};",[],[112,106],"3fb69a02-ef91-468c-81b8-6975649ab16b",["D",1782761585368],"import type { AssistantThreadStartedEvent } from \"npm:@slack/web-api@7.14.1\";\nimport { client } from \"./slack-utils.ts\";\n\nexport async function handleThreadStarted(\n event: AssistantThreadStartedEvent,\n) {\n const { channel_id, thread_ts } = event.assistant_thread;\n console.log(`Thread started: ${channel_id} ${thread_ts}`);\n console.log(JSON.stringify(event));\n\n await client.chat.postMessage({\n channel: channel_id,\n thread_ts: thread_ts,\n text:\n \"Hello, I'm an AI agent powered by Val Town\",\n });\n\n await client.assistant.threads.setSuggestedPrompts({\n channel_id: channel_id,\n thread_ts: thread_ts,\n prompts: [\n {\n title: \"How does persistence work in Val Town?\",\n message: \"How does persistence work in Val Town?\",\n },\n {\n title: \"What can I build on Val Town?\",\n message: \"What can I build on Val Town?\",\n },\n {\n title: \"What's new in Val Town?\",\n message: \"What's new in Val Town?\",\n },\n ],\n });\n}",[],[112,100],"35da5e07-d509-46f4-970c-43b96f524257",["D",1782761585368],"import type { GenericMessageEvent } from \"npm:@slack/web-api@7.14.1\";\nimport { client, getThread, updateStatus } from \"./slack-utils.ts\";\nimport { generateResponse } from \"./generate-response.ts\";\n\nexport async function handleMessage(\n event: GenericMessageEvent,\n botUserId: string,\n) {\n console.log(`Message: ${JSON.stringify(event)}`);\n\n if (\n event.bot_id ||\n event.bot_id === botUserId ||\n event.bot_profile ||\n !event.thread_ts\n ) {\n return;\n }\n\n const { thread_ts, channel } = event;\n await updateStatus({ channel, thread_ts, status: \"is thinking...\" });\n\n // TODO: set a dynamic title after the first user message\n // await updateTitle({ channel, thread_ts, title: \"TODO\" });\n\n const messages = await getThread(channel, thread_ts, botUserId);\n const result = await generateResponse(messages);\n\n await client.chat.postMessage({\n channel: channel,\n thread_ts: thread_ts,\n text: result,\n unfurl_links: false,\n blocks: [\n {\n type: \"section\",\n text: {\n type: \"mrkdwn\",\n text: result,\n },\n },\n ],\n });\n}",[],[112,94],"3222bef9-8481-454e-a06b-8b736d99d011",["D",1782761585368],"import { AppMentionEvent } from \"npm:@slack/web-api@7.14.1\";\nimport { client, getThread, updateStatus } from \"./slack-utils.ts\";\nimport { generateResponse } from \"./generate-response.ts\";\n\nexport async function handleAppMention(\n event: AppMentionEvent,\n botUserId: string,\n) {\n console.log(`App mention: ${JSON.stringify(event)}`);\n\n if (event.bot_id || event.bot_id === botUserId || event.bot_profile) {\n console.log(\"Skipping app mention\");\n return;\n }\n\n const { thread_ts, channel } = event;\n const threadTs = thread_ts ?? event.ts;\n await updateStatus({\n thread_ts: threadTs,\n channel,\n status: \"is thinking...\",\n });\n\n const messages = thread_ts\n ? await getThread(channel, thread_ts, botUserId)\n : [{ role: \"user\" as const, content: event.text }];\n\n const result = await generateResponse(messages);\n\n await client.chat.postMessage({\n channel,\n thread_ts: threadTs,\n text: result,\n unfurl_links: false,\n blocks: [\n {\n type: \"section\",\n text: {\n type: \"mrkdwn\",\n text: result,\n },\n },\n ],\n });\n}",[],[112,88],"9f6a198a-c179-4cc2-a17b-6fafcfe388b9",["D",1782761585368],"import { generateText, ModelMessage, stepCountIs } from \"npm:ai@6.0.92\";\nimport { anthropic } from \"npm:@ai-sdk/anthropic@3.0.45\";\nimport { readFile } from \"https://esm.town/v/std/utils/index.ts\";\n\nconst prompt = await readFile(\"/prompt.md\");\n\nexport const generateResponse = async (messages: ModelMessage[]) => {\n console.log(`generating LLM response...`);\n\n const { text } = await generateText({\n model: anthropic(\"claude-sonnet-4-6\"),\n tools: {\n webSearch: anthropic.tools.webSearch_20250305({ maxUses: 3 }),\n },\n stopWhen: stepCountIs(5),\n system: prompt,\n messages,\n });\n\n console.log(`generated LLM response: ${text.slice(0, 25)}...`);\n\n // Convert markdown to Slack mrkdwn format\n return text.replace(/\\[(.*?)\\]\$(.*?)\$/g, \"<$2|$1>\").replace(/\\*\\*/g, \"*\");\n};",[],[112,76],["D",1782761585368],[251,252,253,254,255],[112],["D",1782761585368],{"_11":12,"_183":82},{"_11":12,"_178":179,"_81":82,"_25":80,"_180":181,"_182":24},{"_33":34,"_13":16,"_27":30,"_167":132,"_168":71,"_169":170,"_171":136,"_172":173,"_174":175,"_176":14,"_177":12},"ctx",{"_46":-5,"_205":334,"_335":336,"_337":338,"_339":24,"_340":24,"_341":24,"_342":144,"_343":-5,"_344":-7,"_345":-7,"_346":20,"_347":24,"_348":24,"_349":-5,"_350":24,"_351":24,"_352":24,"_353":-5},"ENV",{"_305":306,"_307":308,"_309":16,"_310":311,"_312":313,"_314":315,"_316":317,"_318":319,"_320":321,"_322":323,"_324":325,"_326":327,"_328":329,"_330":331,"_332":333},"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"]

History