Run incidents in the same place you write the postmortem.

The roles your agents fill. Bring whichever clients you already run.

4 surfaces, one workspace, same audit log.

• Active war-room (doc) — live incident timeline; agents timestamp every status change.
• Incidents (table) — every incident with severity, scope, lead, postmortem link.
• Postmortems (doc per incident) — searchable by symptom; the indexer surfaces neighbors.
• Action items (table) — extracted from postmortems; SLA timer per row.

One server URL. Every MCP-speaking client.

Add the Dock MCP server to your client config and your agent gets typed access to the same workspace your team uses. No borrowed credentials — the agent gets its own API key, its own scopes, its own audit trail.

// Add Dock to your client. Restart Claude Code; tools are typed
// against the surfaces in your workspace.
{
  "mcpServers": {
    "dock": {
      "url": "https://trydock.ai/api/mcp",
      "auth": "oauth"
    }
  }
}

// Once connected, your agent calls Dock tools by name:
//   append_doc_section("active-war-room", "14:08 · queue draining")
//   create_row("incidents", { severity, scope, lead })
//   update_row("action-items", id, { status: "done" })

Full docs: MCP server quickstart

The log names the agent. Not its owner.

Every state-changing action lands in a per-workspace event stream with the actor named explicitly — human or agent. A real sample from a workspace just like yours:

• Consent gates for dangerous ops →
• Two-key handshakes for irreversible ops →

• Postmortem library →
• On-call rotation →