Cleric is an AI agent tool for SRE (Site Reliability Engineering) and observability scenarios, designed to help engineering teams automatically validate changes in production environments, discover and fix regressions, and thereby reduce customer-visible incidents.
Purpose
The core purpose of Cleric is to serve as a "change gatekeeper" and "on-call assistant" for production environments. After code is merged, it continuously tracks each change, validates its effect under real production traffic, investigates the cause of failures, submits fixes, and verifies recovery after the fix is deployed. Its goal is to let teams discover and fix regressions before customers notice the problem.
Main capabilities
- Change validation: Cleric checks every production change and validates its effect under real traffic. If it discovers a regression (for example, increased latency or a rising error rate), it automatically opens a fix PR and verifies recovery after the fix is deployed.
- Alert handling: Cleric can take over the pager, handle alerts, and close false positives, escalating only real issues that require human intervention, along with investigation findings.
- Production-grade investigation: When an incident occurs, Cleric confirms whether the problem is real, diagnoses the root cause (for example, a configuration change causing connection pool exhaustion), and generates a fix PR for approval, then monitors the fix until recovery.
- Cross-system tracing: Cleric maps services, dependencies, and ownership to trace regressions across systems.
- Continuous learning: Cleric learns from each fix and brings context into future investigations.
How to use
Cleric is designed to integrate quickly into existing technology stacks, and connection can usually be completed within an afternoon. It uses read-only access by default, and write permissions can be enabled when the team is ready. After code is merged, it reads PR information and continuously checks the effect of changes over the following 14 days (for example, the first check after deployment on day 1, weekday peak traffic on day 3, a full sales cycle on day 7, and window closure with no deviation on day 14). When it discovers an anomaly, Cleric sends alerts through app notifications (such as the Cleric APP) and automatically opens a fix PR, waiting for human approval before deployment and verification.
Applicable scenarios
- Engineering teams want to automatically validate every change in production environments and prevent regressions from turning into incidents.
- SRE teams want to reduce alert noise and handle only real issues.
- Scenarios that require quickly locating the root cause of production incidents and obtaining fix recommendations.
- Teams that want to shift part of the on-call burden to AI agents.## Limitations and security Cleric is read-only by default, all operations are logged, and every investigation is auditable. It complies with SOC 2 Type II and undergoes regular manual penetration testing. Data is encrypted at all stages and is not used for training.
Other information
Cleric was named a Cool Vendor by Gartner in the field of AI for SRE and observability. The official page also includes expert podcast clips and customer testimonials, but does not provide specific pricing, a platform support list, or detailed installation steps.
