Runbooks that actually run.
Manifestly turns your incident runbooks, release procedures, and on-call handoffs into scheduled, assignable, auditable workflows. Driven by API and MCP, not tribal knowledge.
MCP and API first · Webhooks and CI hooks · Full audit trail
Trusted in regulated environments, by teams in 25+ countries
SOC 2 Type II compliant · Audit logs on every run
Your infra is automated. Your operational procedures are not.
Critical procedures live in stale wikis and people's heads. Under pressure, nothing enforces the steps, and the post-mortem has no record of who did what.
Procedures go stale
The real runbook lives in someone's head or a wiki page nobody has updated since the last reorg.
Nothing enforces the steps
Under pressure, steps get skipped. A doc cannot require a sign-off or block until a check is done.
No record for the post-mortem
When the review asks who did what and when, the answer is scattered across Slack threads and memory.
Six eng-ops workflows, running on day one.
Concrete procedures your team already needs, turned into workflows that schedule, assign, and audit themselves.
Incident response runbooks
Consistent execution under pressure, with conditional steps by severity and a clean timeline for the review.
Release and deployment
Pre-flight, deploy, and verify, with sign-offs at each gate and a record of every release.
On-call handoffs
Recurring scheduled runs so context transfers cleanly between shifts and nothing drops.
Change management
Approval as a step, with an immutable audit record of who approved what and when.
Access provisioning
Onboarding and offboarding engineers, with security reviews built into the flow.
Recurring maintenance
Cert rotations, backup verification, and scheduled audits that never slip off the calendar.
Engineers do not want another UI. They want it in the pipeline.
Trigger runs, complete steps, and read state programmatically. Manifestly exposes an MCP server and a full API, so your workflows live where your automation already does.
- MCP server so agents and assistants can drive runs directly
- Fires from your deploy pipeline with webhooks and GitHub or CI hooks
- Data collection and Data Tables capture run metadata for reporting
# MCP tool call: start a release runbook
POST /api/v2/runs
{
"workflow": "release-deploy",
"assignee": "on-call@acme.dev",
"trigger": "ci.deploy.success",
"data": { "sha": "a1b2c3d", "env": "prod" }
}
# => run created, steps assigned,
# => webhook fires back on completionHow it works.
Build the workflow once
Turn a runbook or procedure into a reusable workflow with conditional steps, roles, and data fields.
Schedule or trigger it
Run it on a cron, from the API, via webhook from your pipeline, or manually when the page goes off.
Track, assign, and audit
Every run is assigned, tracked to completion, and recorded with a full timeline for the review.
The auditable promise a regulated shop needs.
Every run produces evidence. Role-based assignment, SSO, and SOC 2 Type II back it up.
See our security postureWhy not just a Slack thread or a wiki page?
The defaults work until they do not. Here is what breaks at scale, and what Manifestly adds.
| The default | What breaks at scale | What Manifestly adds |
|---|---|---|
| Slack threads | No structure, no enforcement, scrolls away after the incident. | Enforced steps and a permanent record |
| Confluence or wiki | Goes stale, nobody runs it, no assignment or tracking. | Live runs with owners and due dates |
| A Google Doc | No scheduling, no audit, copy-pasted per incident. | Scheduling, triggers, and an audit trail |
| A homegrown script | One owner, no human-in-the-loop steps, brittle to maintain. | Human and automated steps in one flow |
Make every procedure repeatable, every run accountable.
Workflow software for engineering operations, driven by the API and MCP your team already builds on.
Free limited users