CloudZA® · Interactive capability map
Everything a coding agent does, as one loop you can click through.
Codex reads a repository, plans the work, edits files, runs the result, and reviews its own diff — then loops until the task is done. This map breaks that loop into stages and the stages into concrete capabilities.
The agent loop
Five stages, repeated until the task converges.
Select a stage to filter the capability map to the work that touches it. Most capabilities span several stages; few live in just one.
Capability matrix
What the agent can be asked to do.
Filter by category or loop stage, then open any card for a sample prompt, the output you'd get back, and where it sits in the loop.
Use-case matrix
How real tasks distribute across the loop.
Each row is a task an evaluator might run. The dots show how heavily each stage is exercised — solid for primary, faint for partial.
| Scenario | Understand | Plan | Edit | Run | Review | Primary capability |
|---|
Point it at a repository and watch the loop run.
Codex works where your code already lives — terminal, editor, and CI. Start with a read-only task and approve every diff before it lands.