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",[155],"412f1528-498e-11f1-9272-42b51c65c3df","README.md","019dff15-3d47-712d-aef1-5fa97593c798",["D",1778101075191],"# ai-slackbot\n\nAn AI agent powered by Claude with web\nsearch, 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. **Remix this val** if you haven't already.\n2. **Open your val's HTTP endpoint in a new tab** (you should see it in the homepage of your remixed val),\n then click **Create Slack App**. The app's [manifest](./events.ts) — scopes,\n events, and webhook URL — is pre-filled, so just click **Next** → pick\n 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\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`).\n> Without channel membership, the bot receives the mention event but can't\n> read thread 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 → `lib/handle-thread-started.ts`\n- **Change Slack scopes, events, or app metadata** → edit the `manifest`\n object at the top of `events.ts`. Re-create the Slack app from the updated\n setup 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), both\nadapted for Val Town.",[161],"412f0da8-498e-11f1-9272-42b51c65c3df","lib","019dff15-3d47-712d-aef1-5aacb26c891b",["D",1778101075191],"directory",[167],"pulls","branches","forks","envVars",{"_179":181,"_182":181},"connections",{"_179":20,"_180":20},"count","unfilledCount",3,"emptyCount",{"_32":166,"_13":167,"_142":168,"_144":-5,"_145":55,"_146":22,"_147":78,"_34":238,"_40":20,"_149":170,"_150":-5,"_63":26,"_152":-5,"_193":239,"_108":240},{"_32":231,"_13":232,"_142":233,"_144":168,"_145":55,"_146":22,"_147":78,"_34":234,"_40":20,"_149":206,"_150":235,"_63":26,"_152":-5,"_193":236,"_108":237},{"_32":224,"_13":225,"_142":226,"_144":168,"_145":55,"_146":22,"_147":78,"_34":227,"_40":20,"_149":206,"_150":228,"_63":26,"_152":-5,"_193":229,"_108":230},{"_32":217,"_13":218,"_142":219,"_144":168,"_145":55,"_146":22,"_147":78,"_34":220,"_40":20,"_149":206,"_150":221,"_63":26,"_152":-5,"_193":222,"_108":223},{"_32":210,"_13":211,"_142":212,"_144":168,"_145":55,"_146":22,"_147":78,"_34":213,"_40":20,"_149":206,"_150":214,"_63":26,"_152":-5,"_193":215,"_108":216},{"_32":202,"_13":203,"_142":204,"_144":168,"_145":55,"_146":22,"_147":78,"_34":205,"_40":20,"_149":206,"_150":207,"_63":26,"_152":-5,"_193":208,"_108":209},{"_32":160,"_13":161,"_142":162,"_144":-5,"_145":55,"_146":22,"_147":78,"_34":199,"_40":20,"_149":120,"_150":164,"_63":26,"_152":-5,"_193":200,"_108":201},{"_32":154,"_13":155,"_142":76,"_144":-5,"_145":55,"_146":22,"_147":78,"_34":196,"_40":20,"_149":157,"_150":158,"_63":26,"_152":-5,"_193":197,"_108":198},{"_32":140,"_13":141,"_142":143,"_144":-5,"_145":55,"_146":22,"_147":78,"_34":192,"_40":20,"_149":120,"_150":151,"_63":26,"_152":-5,"_193":194,"_108":195},["D",1778101075191],"children",[],[141],["D",1778101075191],[],[155],["D",1778101075191],[],[161],"412eeb66-498e-11f1-9272-42b51c65c3df","slack-utils.ts","019dff15-3d47-712d-aef1-55f53cedc810",["D",1778101075191],"script","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};",[],[167,203],"412ee2ec-498e-11f1-9272-42b51c65c3df","handle-thread-started.ts","019dff15-3d47-712d-aef1-507d4c5a3816",["D",1778101075191],"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}",[],[167,211],"412ed73e-498e-11f1-9272-42b51c65c3df","handle-message.ts","019dff15-3d47-712d-aef1-4f8af75b2988",["D",1778101075191],"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}",[],[167,218],"412eccf8-498e-11f1-9272-42b51c65c3df","handle-app-mention.ts","019dff15-3d47-712d-aef1-4b42ad0bb5b9",["D",1778101075191],"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}",[],[167,225],"412e9c10-498e-11f1-9272-42b51c65c3df","generate-response.ts","019dff15-3d47-712d-aef1-45147a4680b7",["D",1778101075191],"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};",[],[167,232],["D",1778101075191],[184,185,186,187,188],[167],["D",1778101075191],{"_11":12,"_107":103},{"_11":12,"_100":101,"_102":103,"_73":74,"_104":105,"_106":26},{"_40":22,"_13":16,"_32":55,"_38":56,"_42":33,"_43":57,"_58":49,"_59":39,"_60":52,"_53":14,"_50":12},"ctx",{"_95":-5,"_130":278,"_279":280,"_281":282,"_283":26,"_284":26,"_285":26,"_286":78,"_287":-5,"_288":-7,"_289":-7,"_290":20,"_291":26,"_292":26,"_293":-5,"_294":26,"_295":26,"_296":26,"_297":-5},"ENV",{"_249":250,"_251":252,"_253":16,"_254":255,"_256":257,"_258":259,"_260":261,"_262":263,"_264":265,"_266":267,"_268":269,"_270":271,"_272":273,"_274":275,"_276":277},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","a3e067d4463eea85f6334f8e583e8516ae58c4f6","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"]

Compare

Mergeai-slackbot4/mainintoai-slackbot/main