For platform and SRE teams

Executable runbooks for teams that run production

Runspace puts instructions, checklists and commands in one workspace your team runs together, with your agent alongside, approval on risky steps, and a record of every run.

A Runspace runbook step with its command, approval control and streamed output

Most runbooks are still a wiki page and a terminal

Stale pages

The runbook was right when someone wrote it. Nobody knows if it still is, or which version the on-call engineer followed.

Copy, paste, hope

Commands get copied from a page into a terminal by hand. One wrong paste on a production box is all it takes.

No record of what ran

After an incident, the real fix lives in one person's shell history. Who ran what, and who said yes, is reconstructed from memory.

One workspace, from incident to runbook

Executable runbooks

Instructions, checklists and commands in one workspace

Each step runs inline and its output streams right underneath, like a notebook. Your agent is there along the way, suggesting the next command, and nothing runs without approval.

  • Run steps inline with streamed output
  • Checklists and notes beside the commands
  • Ask your agent (Claude Code or Codex) for help
  • Variables captured from earlier output
A Runspace runbook with Claude Code suggesting the next command inline
Incident workspace

Work the incident in one place, then keep what worked

Capture findings, commands and draft instructions as you go, and request approvals in real time. Afterwards, the incident becomes a runbook draft your team reviews and approves.

  • Capture findings and evidence while you investigate
  • Turn terminal history into runbook steps
  • Request approval mid-incident
  • Review only what changed, then publish
A Runspace investigation panel listing evidence, possible causes and a proposed next check

Built for enterprise operations

Approval pinned to the command

Each run is approved by a separate reviewer, tied to the exact command and runbook revision.

Runbook review

Changes are reviewed and approved before they publish. Published revisions cannot change.

Role-based access

Roles and restricted runbooks control who can edit, run and approve.

Single sign-on

Sign in through your identity provider.

Audit log

A server-side record of who ran what, where, and who approved it.

Decision intelligence

See the decisions behind every incident: what was tried, what the evidence showed, and who approved each step.

Operations performance

Track how runbooks perform: how often they run, where they fail, and how long resolution takes.

How a pilot works

01

Request a pilot

Leave your work email. We'll reach out to talk through your team, your runbooks and what you need from approval and audit.

02

Bring one real runbook

Start with a runbook your team already uses, or capture one from a recent incident's terminal session.

03

Run it with your team

Share it, set roles and restrictions, require approval on risky steps, and check the audit log afterwards.

Built for teams that run production

Runspace is for heads of platform, SRE and infrastructure at engineering orgs of 100+ people, where many engineers touch production and someone has to answer 'who ran that, and who approved it?'. It isn't trying to replace your incident or paging tools. It replaces the wiki page and the copy-paste.

Questions

Is Runspace available today?

Runspace is in private pilots with a small number of teams. Requesting a pilot puts you in line. We'll contact you about timing and fit.

How is this different from open-source runbook tools?

Open-source tools can run commands from a doc. Runspace adds governance: per-run approval by a separate reviewer, pinned to the exact command and revision, unchangeable published revisions, reviewed changes, and a server-side audit log.

What will it cost?

Current plans are Solo at $20/month and Multiplayer at $40 per user per month ($34 billed annually), with a 2-seat minimum. Pricing isn't final and may change before general availability.

Do you have SOC 2 or other certifications?

Not yet. We won't claim certifications we don't have. If your security review needs something specific, tell us during the pilot conversation.

What integrations are supported?

Agent help works with Claude Code and Codex today. For the rest of your stack, tell us what your team relies on when we talk, and it will shape what we build next.

Stop running production from a wiki page

Request a pilot with your work email and we'll reach out to set it up with your team.