Skip to main content
CloudEval has two terminal modes: an interactive terminal UI for humans and explicit commands for scripts, agents, and CI jobs.
CloudEval terminal UI chat workspace with reasoning progress, project context, model, mode, input controls, and account status

​
When to use the terminal UI

Use the terminal UI when you want to stay in a terminal but still inspect CloudEval interactively. 
cloudeval
cloudeval tui
cloudeval tui --tab reports --project <project-id>
CloudEval terminal UI overview tab showing plan, credits, projects, connections, reports, and activity
The terminal UI is useful for:
  • browsing projects
  • jumping into project reports
  • reviewing chat or help context
  • opening exact app pages from a project-oriented terminal session
Do not parse terminal UI output in automation. Use explicit commands with --format json instead.
CloudEval terminal UI reports tab showing report status, score, attention count, run commands, and project report shortcuts
CloudEval terminal UI billing tab showing plan, credits, billing status, subscription controls, and customer portal actions

​
When to use explicit commands

Use explicit commands when another tool needs predictable output. 
cloudeval status --profile codex --format json
cloudeval projects list --profile codex --format json
cloudeval reports run --project <project-id> --type all --wait --profile codex --format json
cloudeval ask "Summarize the highest-risk findings" --project <project-id> --profile codex --format json --non-interactive
CloudEval CLI terminal output showing version, MCP doctor JSON, toolsets, tools, resources, prompts, and project report status JSON
This mode is better for:
  • CI jobs
  • MCP-adjacent automation
  • shell scripts
  • local agents
  • repeatable report generation
  • saving output to files

​
Output rules

For machine-readable commands:
  • stdout is the data channel
  • stderr is for prompts, progress, warnings, and browser-open messages
  • --output <file> writes durable artifacts when supported
  • --profile <name> keeps defaults scoped to one agent or workspace

​
Good defaults for automation

cloudeval doctor --profile codex --format json
cloudeval capabilities --profile codex --format json
cloudeval config show --profile codex --format json
Use those checks before relying on a local CLI install in automation.

​
Next step

Use CLI command reference for the full command surface, or MCP client setup when an agent should call CloudEval through MCP instead of shell commands.
Last modified on June 22, 2026