> ## Documentation Index
> Fetch the complete documentation index at: https://docs.vijil.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Console CLI

> This guide walks through the `vijil` command-line tool. By the end you will have registered an AI agent, run a trust evaluation against it, and retrieved the results.

## Install Vijil Console CLI

<CodeGroup>
  ```bash title="pip" theme={null}
  pip install vijil-console
  ```

  ```bash title="pipx"  theme={null}
  pipx install vijil-console
  ```
</CodeGroup>

### Verify the Installation

```bash theme={null}
vijil --help
```

## Initialize the CLI

Point the CLI at your Vijil Console deployment:

```bash theme={null}
vijil init --url https://console-api.example.com
```

Or run `vijil init` without arguments to be prompted for the URL. The CLI verifies connectivity before saving.

Configuration is stored in `~/.vijil/config.yaml`.

## Login to Vijil Console

```bash theme={null}
vijil login
```

You will be prompted for your email and password. On success the CLI stores your JWT token and automatically selects your team if you belong to exactly one.

If you belong to multiple teams, select one:

```bash theme={null}
vijil team list
vijil team use <team_id>
```

The active team ID is injected automatically into commands that require it.

## Register an Agent

Create an agent configuration pointing at the AI model you want to evaluate.

To avoid putting secrets directly in your shell history, export your provider API key first:

```bash theme={null}
export OPENAI_API_KEY="sk-..."
vijil agent create \
  --agent-name "My Chat Agent" \
  --model-name "gpt-4" \
  --agent-url "https://api.openai.com/v1/chat/completions" \
  --api-key "$OPENAI_API_KEY"
```

Required flags:

* `--agent-name` — a display name for the agent

* `--model-name` — the model identifier (e.g. `gpt-4`, `claude-sonnet-4-20250514`)

Optional flags:

* `--agent-url` — the endpoint the agent is reachable at

* `--api-key` — API key for the agent's provider

* `--agent-system-prompt` — system prompt the agent uses

The output shows the created agent including its `id`. Copy that ID for the next steps.

## List Agents

```bash theme={null}
vijil agent list
```

Output:

```
ID                                    AGENT_NAME       MODEL_NAME  CREATED_AT
a1b2c3d4-...                          My Chat Agent    gpt-4       1712505600
```

Use `--json` for machine-readable output:

```bash theme={null}
vijil agent list --json
```

## List Available Harnesses

Harnesses are test suites that evaluate different trust dimensions:

```bash theme={null}
vijil harness list
```

Output:

```
NAME       UPDATED_AT
safety     1712505600
ethics     1712505600
privacy    1712505600
security   1712505600
toxicity   1712505600
```

## Run an Evaluation

Start a trust evaluation against your agent:

```bash theme={null}
vijil eval run \
  --agent-id a1b2c3d4-... \
  --harness-names '["safety", "security"]' \
  --sample-size 50
```

Required flags:

* `--agent-id` — UUID of the agent to evaluate

* `--harness-names` — JSON array of Harness names to run

Optional flags:

* `--sample-size` — number of prompts to run (1-1000); omit to run all

* `--harness-type` — `standard` (default) or `custom`

* `--wait` — poll until the evaluation completes instead of returning immediately

### Waiting for Completion

Add `--wait` to block until the evaluation finishes:

```bash theme={null}
vijil eval run \
  --agent-id a1b2c3d4-... \
  --harness-names '["safety", "security"]' \
  --sample-size 50 \
  --wait
```

The CLI polls every 5 seconds and prints status updates as the evaluation progresses.

## Check Evaluation Status

If you did not use `--wait`, check the status manually:

```bash theme={null}
vijil eval status <evaluation_id>
```

Output:

```
evaluation_id:  e5f6a7b8-...
status:         running
scores:         None
created_at:     1712505600
started_at:     1712505610
completed_at:   None
```

When the evaluation completes, scores appear:

```
status:         completed
scores:         {"safety": 0.82, "security": 0.67}
completed_at:   1712506200
```

## View Evaluation Results

Retrieve the detailed results:

```bash theme={null}
vijil eval results-detail <evaluation_id>
```

This returns the full results including per-Harness breakdowns and individual Probe results.

Use `--json` to pipe into other tools:

```bash theme={null}
vijil eval results-detail <evaluation_id> --json | jq '.scores'
```

## Generate a Report

Trigger on-demand report generation for the completed evaluation:

```bash theme={null}
vijil eval report <evaluation_id>
```

To download the generated report as HTML or PDF, use the API directly:

```bash theme={null}
curl -s "$VIJIL_URL/evaluations/<evaluation_id>/html?team_id=$TEAM_ID" \
  -H "Authorization: Bearer $TOKEN" \
  -o report.html
```

## Next Steps

* **Custom Harnesses** — create test suites tailored to your use case:

  ```bash theme={null}
  vijil harness custom-create --name "My Harness" --agent-id <id>
  ```

* **Personas** — list presets and create personas for testing:

  ```bash theme={null}
  vijil persona preset-list
  vijil persona from-preset <preset_id>
  ```

* **Policies** — manage compliance policies and rules:

  ```bash theme={null}
  vijil policy preset-list
  vijil policy from-preset <preset_id>
  ```

* **Dome Guardrails** — configure runtime protection:

  ```bash theme={null}
  vijil dome config-create --agent-id <id>
  ```

* **Red Team Campaigns** — run adversarial attack campaigns:

  ```bash theme={null}
  vijil redteam tools
  vijil redteam run --tool garak --agent-url <url> --purpose "chatbot"
  ```

* **Trust Dashboard** — view aggregated trust metrics:

  ```bash theme={null}
  vijil dashboard show
  ```

Every command supports `--help` for detailed usage and `--json` for machine-readable output.