How the Console and CLI relate

Use this page when you want to understand which Veris surface to reach for. If you’re trying to get set up for the first time, go to Quickstart. If you want to see the product in action with zero setup, go to the console and click an example agent.

The mental model

Reach for each surface when it fits the job:

CLI — integration and automation. Initial integration (install, login, create an environment, push your agent) and anything that runs on a schedule (CI/CD gates, nightly regressions, scripted pipelines).
Console — case-by-case review. Opening a specific scenario, reading a specific simulation transcript, spot-checking what happened on a given run. Visual, one-at-a-time inspection.

The console also has example agents — fully working agents pre-wired into simulated services, runnable in one click. These exist for the “see it work” moment and have no CLI equivalent.

Which surface, when

Doing this	Use this
First time seeing Veris, want to see it work	Console → Try an Example Agent
Pushing your agent, iterating on config, running the loop	CLI
Reading a simulation transcript	Console
Debugging why a simulation failed	Console (for the transcript), CLI (for logs)
Scripting anything — CI, scheduled runs, programmatic pulls	CLI
Showing a report to a teammate	Console (shareable URL)
Comparing two runs side by side	Console (Benchmarks)
Regenerating scenarios, pushing a new image	CLI

Console surfaces ↔ CLI commands

Most console pages have CLI equivalents that work with the same data. Use this table when you see something in the console and want to know how to do the same thing from a script, or vice versa.

Console page	What it shows	CLI equivalents
Overview	Onboarding for new users (“Connect Your Agent” and “Try an Example Agent” side by side); a stats dashboard with recent runs once you have environments.	—
Environments	Environments in the active profile’s org. Clicking one shows its sandbox config, env variables, and image tags (each with a last-pushed timestamp).	`veris env create`, `veris env push`, `veris env list`
Scenarios	Scenario sets and their scenarios, plus the grader mapped to each set.	`veris scenarios create`, `veris scenarios list`, `veris scenarios status`
Simulations	Simulation runs and their per-simulation transcripts, with evaluation scores and assertion results shown inline once evaluations complete.	`veris simulations create`, `veris run`, `veris evaluations create`
Benchmarks	Side-by-side comparison of runs and nightly-trend charts for an environment.	(console only today)
Reports	Reports with root-cause analysis and suggested fixes. URL is `/agent-audits`.	`veris reports create`, `veris run --report`
Training	Fine-tune jobs. Beta.	(console only today)

What the console does well

Transcript reading. Clicking into a simulation shows the full conversation between the actor and your agent, including the actor’s reasoning and every service API call. This is the single most valuable debugging view Veris provides; the CLI intentionally doesn’t reproduce it.
Shareable URLs. Every environment, run, and report has a stable URL. Paste one into Slack or a PR description.
Real-time monitoring. veris run streams status in the terminal, but a teammate can watch the same run in the console without needing CLI access.

What the CLI does well

Everything scriptable. Pushing new image tags, regenerating scenarios after a code change, gating PRs on eval scores, pulling run data into custom dashboards.
Full pipeline in one command. veris run chains simulations → evaluations → report and returns a markdown summary to stdout — ideal for CI comments.
Non-interactive mode. Every command takes flags so it works headless.

A rule of thumb: if you’d want to do it again next week on a different agent, it belongs in the CLI. If you’d want to look at it with someone else, it belongs in the console.