Open-source status page, incident response and uptime monitoring
TypeScript alertingincident-managementmcpmcp-servermonitoringmonitoring-as-codeobservabilityon-callopen-sourcesrestatus-pagestatuspagesynthetic-monitoringterraformtinybirdtursouptimeuptime-checkeruptime-monitor
import { getLogger } from "@logtape/logtape";
import type { KnownBlock, WebClient } from "@slack/web-api";
import { redis } from "@/libs/clients";
import { DOCS_URL } from "./home";
const logger = getLogger("api-server");
// Suggested prompts live in `features.agent_view.suggested_prompts` in
// slack-manifest.json. The agent experience renders them at the top of the
// Messages tab rather than inside a thread, so there is no runtime call to
// make and no second copy here to drift from the manifest.
/**
* First contact. Short, because it sits above the manifest's prompt chips,
* which already show what to ask for. It exists to say the one thing nobody
* should have to discover by trying it: the agent drafts, and nothing reaches
* the status page without a click.
*/
const WELCOME_TEXT =
"I turn incident conversations into status page updates. I draft, you approve — nothing is published until you click Approve.";
const WELCOME_BLOCKS: KnownBlock[] = [
{
type: "section",
text: {
type: "mrkdwn",
text: "I turn incident conversations into status page updates.\n\nTell me what's happening, or mention *@openstatus* in a channel and I'll read the thread. *I draft, you approve* — nothing reaches your status page until you click *Approve*. I can also answer questions about your monitors and the openstatus docs.",
},
},
{
type: "context",
elements: [
{
type: "mrkdwn",
text: `Drafts are AI-generated — read them before approving. <${DOCS_URL}|Documentation>`,
},
],
},
];
/**
* Whether this person has met the agent before. The agent experience has no
* per-thread empty state to greet into — opening the Messages tab is the only
* hook, and it fires every time — so the greeting is once per person.
*/
const GREETED_PREFIX = "slack:greeted:";
const GREETED_TTL_SECONDS = 365 * 24 * 60 * 60;
function greetedKey(teamId: string, userId: string): string {
return `${GREETED_PREFIX}${teamId}:${userId}`;
}
/**
* Greets someone the first time they open the agent, and never again.
*
* `threadTs` is set only on the legacy `assistant_thread_started` path, where
* the greeting belongs in the thread that was just opened.
*/
export async function greetOnce(args: {
slack: WebClient;
teamId: string;
userId: string;
channel: string;
threadTs?: string;
}): Promise<void> {
const { slack, teamId, userId, channel, threadTs } = args;
// Claimed before posting, not after: opening the Messages tab twice fires two
// events, and a check-then-set greets on both. Released if the post fails.
const claimed = await redis.set(greetedKey(teamId, userId), "1", {
nx: true,
ex: GREETED_TTL_SECONDS,
});
if (!claimed) return;
try {
await slack.chat.postMessage({
channel,
...(threadTs ? { thread_ts: threadTs } : {}),
text: WELCOME_TEXT,
blocks: WELCOME_BLOCKS,
});
} catch (err) {
await redis.del(greetedKey(teamId, userId));
throw err;
}
logger.info("slack greeted user", { teamId, userId, channel });
}
// Slack clears the status as soon as the app posts in the thread, so there is
// no matching "clear" call on the success path. Superseded by
// `setSessionStatus` — kept for workspaces the agent session API rejects.
export async function setAssistantStatus(
slack: WebClient,
channel: string,
threadTs: string,
status: string,
): Promise<void> {
await slack.assistant.threads.setStatus({
channel_id: channel,
thread_ts: threadTs,
status,
});
}
// Agent session lifecycle: `processing` shows the agent as working on the
// thread, `active` hands the turn back to the user. Works on thread-based
// sessions in channels and DMs; throws `feature_disabled` on workspaces where
// agent sessions aren't enabled.
export async function setSessionStatus(
slack: WebClient,
channel: string,
threadTs: string,
status: "processing" | "active",
initiatorUserId?: string,
): Promise<void> {
await slack.agents.sessions.setStatus({
channel_id: channel,
thread_ts: threadTs,
status,
initiator_user_id: initiatorUserId,
});
}