Run AI agents in production with confidence. CompTIA AgentOps+ validates your skills to deploy, govern, monitor, and troubleshoot AI agent systems across enterprise environments. This vendor-neutral expansion certification is built for experienced DevOps, cloud, and IT operations professionals who keep agentic AI reliable, secure, and cost-effective at scale. CompTIA AgentOps+ launches in Q2 2027, along with CompTIA CertMaster learning products to help you build these skills.
V1 — Coming Q2 2027
AgentOps+ (Coming Q2 2027)
Skills you'll learn
Build the skills to run AI agents in production with CompTIA learning and validate them with AgentOps+ certification.
Design single-agent and multiagent systems, apply reasoning patterns like ReAct and Chain-of-Thought, and deploy agents across containerized, serverless, cloud, and on-premises environments.
Connect agents to enterprise tools through MCP servers and OpenAPI definitions, and build RAG pipelines and knowledge bases that keep agent responses grounded.
Enforce policy as code, guardrails, human-in-the-loop checkpoints, and emergency stop controls so agents act within organizational policy.
Secure agent identities, secrets, and tool permissions, and apply data protection controls aligned to the OWASP LLM and Agentic Top 10 and MITRE ATLAS.
Trace agent behavior with OpenTelemetry, set SLO-based alerts, control token spend with budgets and circuit breakers, and resolve deployment, behavioral, and performance issues.
Ship agents through CI/CD with evaluation and safety gates, detect drift with continuous testing, and respond fast when agents misbehave or go down.
Exam details
Exam version: V1
Exam series code: AG0-001
Launch date: Q2 2027
Languages: English
Recommended experience: 3–4 years in a DevOps, cloud, systems administration, network engineering, or IT administration role (CompTIA Cloud+ or Linux+ recommended), some programming or scripting experience, and 6–12 months of exposure to AI/ML concepts
Build your AgentOps skills with CompTIA CertMaster
CompTIA is creating a suite of CertMaster learning products for AgentOps+, available in Q2 2027. Choose CertMaster Perform for instruction, hands-on labs, and assessments in one complete experience, or CertMaster Labs for focused, hands-on practice on real machines. Either way, you’ll build practical skills for running AI agents in production as you learn what’s covered on the AgentOps+ exam.
| Perform | Labs | |
| Best for: |
Best for those looking to build skills, learn concepts, and gain hands-on experience. |
Best for those looking to gain hands-on experience applying skills. |
| Primary purpose: |
Comprehensive learning with robust set of lab activities in real and simulated environments to practice skills and job readiness. |
Apply skills in real-world scenarios. |
| Contains: |
Instructional content, video, interactives, labs (simulated and live virtual machines), assessments, practice tests |
Live virtual lab environment with guided tasks and real world-scenarios |
| Estimated duration: |
30–60 hours |
15–25 hours |
AgentOps+ (V1) exam objectives
AI Agent Architecture, Deployment, and Configuration (30%)
- Explain architectural components and design patterns for AI agent systems: Compare single-agent and multiagent (collaborative, hierarchical, swarm) systems, reasoning patterns (ReAct, Chain-of-Thought, Tree-of-Thought, state machine), memory tiers, API-based vs. self-hosted models, and autonomy levels.
- Configure AI agent deployment and orchestration workflows: Set up Agent-to-Agent (A2A) coordination and handoffs, task delegation, containerized and serverless deployments, cloud vs. on-premises hosting, and autoscaling.
- Implement tool integrations for AI agents: Install and configure MCP servers and gateways, define OpenAPI-based tools, connect databases and vector stores, and manage session state and context windows.
- Configure and manage agent prompt templates, instructions, and configuration systems: Version system instructions, route prompts, apply declarative and environment-specific configurations, and enforce structured outputs and response validation.
- Implement knowledge bases and RAG pipelines: Run ingestion, chunking, embedding, retrieval, and reranking; manage vector indexes; apply hybrid, agentic, and Graph RAG; and track citations to reduce hallucinations.
Agent Governance, Security, and Compliance (22%)
- Implement governance policies for AI agent operations: Apply policy as code, pre-execution validation of tool calls, privileged and just-in-time access, segregation of duties, agent ownership and risk tiers, and human-in-the-loop policies.
- Implement guardrails for AI agent operations: Use approved tool catalogs and allow lists, parameter validation, scope limits, input and output guardrails, and emergency stop and override mechanisms.
- Configure security controls for AI agent authentication, authorization, and secrets management: Secure agent identity with service accounts, RBAC, least privilege, and SPIFFE/SPIRE; manage secrets and API keys; and scope OAuth 2.0 and MCP server permissions.
- Implement data governance and privacy controls for AI agents: Detect and redact PII, minimize data in prompts and memory, enforce trust boundaries, and generate audit evidence aligned to the OWASP LLM Top 10, OWASP Agentic Top 10, and MITRE ATLAS.
Agent Observability, Performance, and Cost Management (23%)
- Configure observability instrumentation for AI agent systems: Implement structured, prompt, and tool call logging; OpenTelemetry distributed tracing with spans for each reasoning step; and metrics for token usage, latency, cost, and success rates.
- Configure alerting and monitoring for agent execution, output validation, and behavioral anomalies: Validate tool invocations and structured outputs, detect unexpected tool use, set threshold and baseline anomaly alerts, and track SLAs, SLOs, and SLIs.
- Implement cost monitoring and optimization: Track token consumption and LLM API spend; set token budgets, cost ceilings, and circuit breakers; and optimize with model right-sizing and semantic caching.
- Troubleshoot AI agent deployment and integration issues: Diagnose model API connectivity, authentication, resource, tool integration, and state persistence failures, then resolve them with rollbacks and configuration drift remediation.
- Analyze and troubleshoot AI agent behavioral and performance issues: Identify hallucinated tool calls, reasoning loops, intent drift, and multiagent handoff failures; measure task completion and trajectory efficiency; and remediate with prompt, tool, model, and context changes.
Agent Operations (25%)
- Implement and operate CI/CD pipelines for agent systems: Version each agent as a single deployable unit (model configuration, prompts, tool and MCP definitions, policies, and memory); add evaluation, safety, and vulnerability gates; and promote with GitOps, IaC, canary, and blue-green rollouts.
- Manage agent connectivity and rate controls: Apply agent-to-tool network policies, private endpoints, and Zero Trust principles, plus per-agent rate limiting, circuit breakers, and retry policies.
- Operate continuous evaluation, testing, and validation: Run smoke, regression, and golden dataset evaluations; monitor drift; and test against prompt injection, jailbreaks, tool manipulation, data exfiltration, and memory poisoning.
- Perform incident response for agent misbehavior and outages: Detect harmful output and unauthorized actions, activate emergency stops, revoke credentials, conduct root cause analysis, and restore service against RTO/RPO targets, including cascading multiagent failures.
Preview the full AgentOps+ V1 exam objectives (draft version) here!