
Heystack
dev.heystackPublished by heystack.devOfficial publisherv0.1.0Updated Oct 24, 2026
Observability for AI apps: investigate traces, logs, LLM usage, replays and crashes; manage alerts.
Overview
Heystack's hosted MCP server lets an assistant investigate AI app traces, logs, crashes and LLM usage, and manage apps, alerts and workspace settings.
- What you can do
- Check app health and open a one-call incident dossier for an investigation
- Search traces and logs in plain English, open and explain traces, session stories and replays
- Review issues and crashes with the suspect release, plus gen-AI usage and cost
- Manage apps, ingest keys, sampling and replay configuration, alert rules and app care
- Opt-in toolsets add team administration (members, invites, access tokens, audit) and read-only SQL
- When to use it
- Useful when you want a coding agent to debug why an endpoint or feature is failing after a deploy, review LLM usage and cost, or set up and verify Heystack instrumentation in a project without leaving the editor.
- Requirements
- Remote streamable HTTP endpoint at mcp.heystack.dev; no local package. Sign-in is OAuth with Heystack (choose an organization); alternatively an Authorization header carrying a Heystack personal access token. Team and SQL toolsets require editing the connection URL.
Installation
In SourceWeft
- Open Heystack in the dashboard and add it to a workspace.
- Connect the required credentials. They are stored privately and never shown on public pages.
- Enable the server for the chats that should use its tools.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Use with an agent
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/dev.heystack%2Fheystack/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Other MCP clients
Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.
{
"mcpServers": {
"heystack": {
"type": "http",
"url": "https://mcp.heystack.dev/mcp"
}
}
}README
Heystack plugins
Plugins for Heystack — observability + security for AI apps — so your coding agent can set up Heystack, investigate incidents, and manage your workspace from your editor.
Install (Claude Code)
Then run /mcp to sign in with Heystack and pick your organization. Try: "list my Heystack apps", "why is /checkout failing since the last deploy?", or "set up Heystack in this project and verify it's receiving data".
What you get
heystack (recommended) — the all-in-one plugin:
- Hosted MCP server at
https://mcp.heystack.dev/mcp(connected with?toolsets=core,manage): app health, search traces & logs in plain English, open and explain traces, session stories and replays, issues and crashes with the suspect release, gen-AI usage & cost, a one-call incident dossier (heystack_investigate); plus apps, ingest keys, sampling/replay config, alert rules and app care. Team administration (members, invites, access tokens, audit) and read-only SQL stay opt-in: edit the plugin's.mcp.jsonURL to?toolsets=core,manage,team(oradvanced). Actions that grant access, mint keys or tokens, or send data to an external URL always ask for your confirmation. Sign in with Heystack (OAuth) — no keys to paste. Full reference: https://heystack.dev/docs/mcp heystack-setupskill — detects the runtime (Next.js, Cloudflare Workers/Agents, Node, browser, Android) and wires the correct@heystack/otelentry or OTLP path; with the MCP server it creates the app and key itself and verifies data arrives withheystack_verify_setup.heystack-investigateskill — the investigation loop: overview → dossier → traces/sessions → fix in code → verify; when to change issue status or create alerts; never destructive without your confirmation.
heystack-setup — the setup skill on its own (no MCP server), kept for existing installs: /plugin install heystack-setup@heystack.
heystack-timeline (optional, requires heystack) — /plugin install heystack-timeline@heystack. Hooks that show your coding agent's steps on the Heystack investigation timeline.
- Opt-in: nothing is recorded until you switch on Show my coding-agent steps in the Heystack console (Settings, Profile). Turning it off deletes the recorded steps.
- Privacy: only the file path of an edit and the one-line description of a shell command are sent. Never file contents, diffs, command lines, prompts or tool output. Absolute paths transit TLS to Heystack and are reduced to repo-relative paths before storage; paths outside the project are dropped.
- Best-effort: Claude Code gives session-end hooks about 1.5 seconds, so the closing call may not land; the investigation then just goes idle. A shell command without a description records no step.
- Requires write access on your Heystack connection; it reuses the
heystackplugin's MCP sign-in.
Other agents
- Cursor / VS Code / Codex / Gemini CLI / Claude.ai connect to the same MCP URL — see https://heystack.dev/docs/mcp for one-click links and commands.
- The skills are plain Agent Skills files and are also served raw at https://heystack.dev/heystack-setup.md and https://heystack.dev/heystack-investigate.md. The root
plugin.json/mcp.json/skills/follow the cross-vendor Agent Plugins layout. - Every docs page is available as Markdown: https://heystack.dev/llms.txt
Layout
Source of truth for the skills is the heystack monorepo (packages/setup-skill, packages/investigate-skill); this repo is a mirror.
Source: README.md at commit d20379f
Tools
0Version history
1- v0.1.0LatestSep 17, 2026
Score
75% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
- Reached in the last 30 daysRequired25 pts
- Auth declaredRequired10 pts
- Tools listed15 pts
Answer tools/list without a credential, or ship a package that starts in a sandbox without one.
- DescriptionRequired15 pts
- Logo10 pts
- Documentation15 pts
- License10 pts
Declare a license in the repository or the registry entry.