Connections and remediation access
The hub for the connection pages: how your estate becomes tools the agents can call, which credential may do what, ticket systems, and one page per connector type.
Concept · Updated 2026-10-23 · Describes release 0.1.164
What these pages are for
An agent can only investigate what it can reach. Datasources is where you connect the systems of your estate — metrics, logs, clusters, clouds, git hosts, ticketing, on-prem hosts and network devices — and each connection becomes a tool the pillar agents can call.
Connect access once. Use separate investigation and execution credentials, or one credential for both purposes. The incident or Change prepares the proposed operation, target and connection for human approval. Supported structured actions need no per-action route setup.
The pages
The chapter that used to live at this address is now the pages below, in the order its sections had.
- Manage connections — the concepts, the connector catalog, the connect form per kind, keyless cloud access, secrets and rotation, Test and its failures, the connected list and its actions, legacy routes, troubleshooting and limits.
- Read and write roles, and the write gate — Use this connection for, splitting a credential, several connections into one cloud account, what counts as a write, the control matrix, and how a prepared change picks its connection.
- Ticket systems — Jira Cloud, ServiceNow and Azure DevOps: one ticket per incident, filed at open, written back and closed at resolve.
Integration pages
One page per connector type: its fields, what Test runs, the tool the agents get and its read surface. Alertmanager, Argo CD, AWS, Azure, Azure DevOps (Repos, Boards), Datadog, Dynatrace, Elasticsearch / OpenSearch, FortiGate (FortiOS REST API), Generic API connector, Generic CLI connector, Generic MCP connector, Gitea, GitHub, GitLab, Google Cloud, Grafana, Grafana Tempo, Jaeger, Jira, Kubecost, Kubernetes (kubectl + kubeconfig), Kubernetes (MCP), Log Analytics workspace (KQL) — optional, Loki, Network device (Cisco / pfSense via netmiko), New Relic, Opsgenie, PagerDuty, Palo Alto Networks (PAN-OS API), PostgreSQL, Prometheus, RabbitMQ, ServiceNow, Splunk, SSH (remote host, read-only), VictoriaMetrics, Zabbix.