Skip to content

Image: static-www.elastic.co · rights & removal

Executive Summary

The solution addresses the challenge of tracking OpenShift compliance across a fleet by centralizing compliance data in Elasticsearch and leveraging Elastic Agent Builder to create interactive AI agents for investigation and action. The core problem stems from compliance results being stored locally on each cluster, making fleet-wide visibility difficult, and manual rule tracking existing separately from scan results.
The ingestion process involves deploying an Elastic Agent on each cluster to poll the OpenShift Compliance Operator APIs every 60 seconds, collecting scan results, check details, and remediation information. This data is ingested into Elasticsearch through a custom integration that parses and normalizes the results, creating dedicated indices for scans, checks, and remediations.
The system enables platform and security teams to query compliance posture in plain English. It facilitates investigating failures by tracing events across historical scan data, correlating with manual rule tracking, and using Elastic Workflows to annotate manual rules based on custom Common Expression Language (CEL) evaluations. The architecture is supported by a structured Agent Builder design involving custom instructions, specialized skills for posture, failure investigation, and action, targeted tools for querying Elasticsearch, and deterministic workflows that interact with the cluster via secure connectors.

Facts Only

* An Elastic Agent polls OpenShift Compliance Operator APIs every 60 seconds on each cluster.
* Scan results and check results are stored locally on each cluster initially.
* Compliance data is ingested into Elasticsearch via a custom integration that queries OpenShift API endpoints.
* The ingestion process uses three transforms: scans, checks, and remedies, collapsing streams into dedicated latest-state indices (ospm-compliance-\*-latest).
* The agent collects data from specific Compliance Operator resources: `compliancescans`, `compliancecheckresults`, and `complianceremediations`.
* Custom rules are tracked via `CustomRule` objects evaluated using CEL expressions.
* A scheduled workflow annotates manual rules based on custom scan statuses ingested into Elasticsearch.
* Elastic Agent Builder is used to create an agent that queries compliance data in natural language.
* Skills include Posture summary, Failure investigation, Live cluster actions, and Reporting.
* Tools include ES|QL queries against the latest-state indices and searches against raw time-series data streams.
* Workflows interact with live cluster state via tools like `ocpcompliance.queryliveworkload` and `ocpcompliance.setremediationstate`, requiring user confirmation.
* The Elastic Agent is deployed as a single replica per OpenShift cluster, injecting the `CLUSTERNAME` environment variable for event tagging.

Full Take

The narrative positions a significant shift from siloed, reactive compliance auditing to an integrated, proactive posture management system. The fundamental pattern being addressed is the operational friction created by distributing critical status information across disparate cluster consoles and spreadsheets, which leads to delayed detection of policy drift and auditor reporting inefficiencies. The proposed solution leverages the strengths of modern data platforms—Elasticsearch for centralized, searchable evidence, and LLM-powered agents (Agent Builder) for contextual reasoning over that data.
The complexity introduced by manual controls, which necessitate external tracking and reconciliation, highlights a gap between automated scanning and human governance. The integration of CEL with Elastic Workflows to bridge this gap represents an attempt to formalize the manual oversight process into an automated feedback loop, allowing risk assessments (via custom rules) to influence real-time operational outcomes (via annotations).
The design choice to rely on tightly controlled, deterministic workflows for live cluster actions, rather than leaving decision-making to the LLM alone, is a critical acknowledgment of the risks inherent in granting AI autonomy over infrastructure. The structure—combining data ingestion pipelines, structured rule evaluation, and human-in-the-loop execution—suggests an emerging pattern where AI functions not as an autonomous actor but as a highly sophisticated orchestrator for evidence correlation and proposing controlled remediation steps grounded in pre-defined governance constraints. The implication is that achieving fleet-scale compliance requires formalizing the relationships between static policy (CEL rules), dynamic state (scan results), and actionable execution (workflows).
Bridge Questions: If a system can automate the tracing of policy drift, what organizational changes are required to shift accountability for the initial definition of those policy rules from platform teams to security/operations teams? What safeguards must be established within the workflow system to prevent automated remediation based on inferred historical context from leading to systemic risk? How does this centralized view change the incentive structure for OpenShift administrators when manual oversight is demonstrably surfaced through an AI interface rather than scattered documentation?

From the original · Elastic Security

When an OpenShift compliance control fails, Elastic Agent Builder traces the root cause, from Compliance Operator scans centralized in Elasticsearch and CEL checks for the rules the scanner marks manual. A single Elastic Agent on each cluster polls OpenShift Compliance Operator APIs every 60 seconds.
Read the full story at elastic.co

Sentinel unavailable

The automated check of the source article's wording did not complete, so no result is shown.

This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.

From posture to root cause: Investigate OpenShift compliance with Elastic Agent Builder | Huntaegis