# Heystack

> Heystack is managed observability + security for AI apps. Apps push OpenTelemetry to Heystack; people read it in the console at https://console.heystack.dev; coding agents read and act on it through the hosted MCP server. Heystack is never in the request path.

Every docs page below is available as raw Markdown at the listed `.md` URL. The full set is concatenated in https://heystack.dev/llms-full.txt.

## For agents

- [MCP server endpoint](https://mcp.heystack.dev/mcp): Streamable HTTP, OAuth 2.1 (sign in with Heystack) or `Authorization: Bearer hs_pat_…`. Toolsets via `?toolsets=core|manage|team|advanced|all`, `?readonly=true`.
- [Connect: Claude Code](https://heystack.dev/docs/mcp.md#claude-code): `claude mcp add --transport http heystack https://mcp.heystack.dev/mcp`
- [Plugin (Claude Code)](https://github.com/heystack-hq/heystack-plugins): `/plugin marketplace add heystack-hq/heystack-plugins` then `/plugin install heystack@heystack` — MCP server + skills.
- [heystack-setup skill](https://heystack.dev/heystack-setup.md): wire the SDK for the detected runtime (Next.js, Cloudflare Workers, Node, browser, Android).
- [heystack-investigate skill](https://heystack.dev/heystack-investigate.md): the investigation loop using the MCP tools.
- [REST API base URL](https://ingest.heystack.dev/api): same personal access tokens; see the REST API page.

## Getting started

- [Introduction](https://heystack.dev/docs/introduction.md): Heystack is managed observability and security for AI apps. You ship your app; Heystack watches it, tells you in plain language when something breaks, and hands you the fix.
- [Quickstart](https://heystack.dev/docs/quickstart.md): Get your AI app reporting to Heystack in a few minutes. You'll create an app, grab an ingest key, wire up the SDK, and watch telemetry land in the console.
- [Core concepts](https://heystack.dev/docs/concepts.md): A quick mental model of how Heystack organizes your data.

## SDK & frameworks

- [The SDK](https://heystack.dev/docs/sdk.md): @heystack/otel is runtime-aware telemetry that exports to Heystack. There's one import per runtime, so the root entry stays safe to load anywhere (including Edge and Workers)…
- [Next.js](https://heystack.dev/docs/nextjs.md): Use the @heystack/otel/next entry for any Next.js deploy target — Vercel/Node and Cloudflare/OpenNext alike. registerHeystack is runtime-aware: it auto-detects Node vs Cloudflare…
- [Cloudflare Workers](https://heystack.dev/docs/cloudflare-workers.md): Heystack uses Cloudflare's native OpenTelemetry export as the primary Workers integration. Cloudflare creates the invocation and platform spans, automatically traces operations…
- [Node.js](https://heystack.dev/docs/nodejs.md): For a long-running Node server — Express, Fastify, NestJS, or plain Node — use the @heystack/otel/node entry. initHeystack bundles auto-instrumentations (HTTP and common…
- [Android](https://heystack.dev/docs/android.md): Heystack accepts Android product telemetry as standard OpenTelemetry logs over OTLP/HTTP. There is no Heystack Android SDK to install: use your app's existing HTTP client and…
- [OpenTelemetry Collector](https://heystack.dev/docs/collector.md): If you already run an OpenTelemetry Collector — or you have non-JavaScript services (Go, Python, Java, Ruby, etc.) that already emit OTLP — you can route telemetry straight to…

## Platform

- [The console](https://heystack.dev/docs/console.md): The Heystack console is where you manage apps, mint ingest keys, and explore your telemetry. All telemetry is scoped per app — you only ever see data for the app you're looking at.
- [Teams & organizations](https://heystack.dev/docs/teams-and-organizations.md): Heystack is built for teams. When you sign up, we automatically create a workspace for you — your own private space where all your apps and their data live.
- [Session Replay](https://heystack.dev/docs/replay.md): See exactly what your users saw. Heystack records real user sessions — clicks, scrolls, page transitions, and on-screen errors — so you can watch a replay when something goes…
- [Crash Reporting](https://heystack.dev/docs/crash-reporting.md): Heystack automatically groups, symbolicates, and surfaces native crashes from your iOS app. When your app crashes, the next launch captures the crash report and sends it to…
- [Linking into the console](https://heystack.dev/docs/deep-links.md): Every view in the Heystack console is addressable by URL, so another tool — an internal ops page, a runbook, an alert message, a chat bot — can drop someone straight onto the…

## Agents & API

- [MCP server](https://heystack.dev/docs/mcp.md): Let your coding agent investigate and fix issues for you. The Heystack MCP server connects Claude Code, Cursor, VS Code, Codex, Gemini CLI, Claude.ai and any other MCP-capable…
- [Agent playbooks](https://heystack.dev/docs/agent-playbooks.md): Copy-paste prompts for any coding agent connected to the Heystack MCP server. Each one names the tools it leans on so you can see what the agent is doing — and so you can hand…
- [REST API & read keys](https://heystack.dev/docs/api.md): Everything the console and the MCP server can do is available as a plain HTTPS API — for health checks, status pages, internal dashboards, CI checks, scripts, and autonomous…

## Optional

- [SDK on npm](https://www.npmjs.com/package/@heystack/otel): `@heystack/otel` — runtime-aware entries `/node`, `/next`, `/workers`, `/web`.
- [Console](https://console.heystack.dev): apps, keys, explorer, Agents & API page (personal access tokens, agent activity).