GAEZLA agents let an operations team run controlled work on enrolled systems without turning every task into a remote desktop session.
From assigned work to a reviewable result
Each agent:
- authenticates with its own credential;
- receives work assigned to that agent and target;
- verifies the workload signature against its trusted signing configuration;
- runs the supported workload locally; and
- returns status, logs, timing, and result metadata to GAEZLA.
The resulting execution record shows the playbook, target, supplied variables, queue state, returned output, and failure context. Operators can review what ran without rebuilding the story from terminal history.
Investigation and execution stay separate
Investigation uses read-only tools to collect evidence. When a case supports an external change, GAEZLA creates a proposal with the evidence, scope, risk, safeguards, stop conditions, and rollback. Approval and any required step-up happen before that proposal can execute.
Playbooks, schedules, and routines use their configured execution controls. Keeping these paths distinct makes it clear whether an operator is reading, approving a proposed change, or reviewing work that an agent returned.