Tools & APIs

The action surface for your stack,and everything that talks to it.

OpsZ exposes a governed execution layer through both a native MCP server and a REST API. Agents, scripts, external tools, and CI/CD pipelines can query inventory, trigger workflows, observe job outcomes, and act on your infrastructure, all within the same policy and audit framework that governs human actions.

Available to early customers and qualified technical evaluators.

Two surfaces, one governed layer

MCP server

For agents and AI-native tools

OpsZ exposes a fully MCP-compliant server. Any compatible agent can connect, discover available tools, query live infrastructure context, and request governed workflow execution through a standard interface. No custom integration required.

REST API

For scripts, pipelines, and existing tooling

A full-coverage REST API for programmatic access to every OpsZ capability. Trigger jobs, query fleet state, manage workflows, read audit logs. Designed to fit into existing automation, CI/CD, and operational tooling without friction.

Same governance, every call

Every API and MCP call is subject to the same RBAC, approval gates, and blast-radius policy as actions taken in the UI. There is no ungoverned path.

Full audit trail

Every call is attributed, timestamped, and logged. Whether the actor is a human, a script, or an AI agent, the record is the same.

Token-scoped access

API tokens are scoped to specific roles and capabilities. Agents get exactly the access they need, nothing more.

Live state, not cached data

Inventory queries return live fleet state from the real-time backplane. Agents reason on current reality, not last night's snapshot.

MCP server

Query, act, observe. Through a standard interface.

The OpsZ MCP server exposes tools across six capability groups. Any MCP-compatible agent can connect and begin querying and acting immediately.

Inventory

Query hosts, clusters, environments, and live topology

8 tools

Workflows

Discover, build, version, and execute workflow templates

14 tools

Jobs

Trigger jobs, observe execution, retrieve logs and outcomes

12 tools

Identity

Manage users, roles, groups, and access requests

10 tools

Secrets

5 tools

Policy

Inspect and manage blast-radius and image allowlist policies

4 tools

Example: agent queries live fleet inventory

# Tool: inventory__query_hosts

tool: inventory__query_hosts
input:
  environment: "production"
  status: "active"
  page_size: 100

# Returns: live host list with OS, agent status, environment, last seen
# All calls governed by RBAC scope and logged to audit trail

Full tool schemas available to early customers on request.

REST API

Programmatic access to every platform capability.

Designed for integration into existing operational tooling, CI/CD pipelines, and custom automation, without replacing what is already working.

GET/v1/inventory/hostsQuery live host inventory with filters

POST/v1/jobsExecute a workflow against a target set

GET/v1/jobs/{id}/logsRetrieve execution logs for a job

GET/v1/workflows/templatesList available workflow templates

POST/v1/workflows/templates/{id}/versionsSnapshot a template to a new version

GET/v1/auditQuery the full audit log with filters

Example: trigger a governed workflow via REST

POST /v1/jobs
Authorization: Bearer {token}

{
  "template_name": "openssl-cve-remediation",
  "environment": "staging",
  "targets": ["prod-web-01", "prod-web-02"],
  "variables": { "dry_run": false }
}

# Returns: job_id, blast_radius_preview, policy_result
# Job blocked if policy conditions not met. Full audit recorded regardless.

Full endpoint reference available to early customers and qualified evaluators.

API access is available by request.

Full documentation, authentication details, and sandbox access are available to early customers and qualified technical evaluators. Tell us about your environment and we will get you set up.

✓ Full MCP tool schema✓ REST API reference✓ Sandbox environment✓ Auth and token setup✓ Direct team access

Available to early customers and qualified technical evaluators. Not publicly available at this time.