Docs / Reference and troubleshooting

Command line tool

The verbiqo command gives you Verbiqo's abilities in a terminal: pipe a log into it, rewrite text in a script, or hold a conversation.

Install it

1. Choose Install Command Line Tool

   Click the Verbiqo icon in the menu bar and choose Install Command Line Tool….

2. Open a new terminal window

   The command is installed in /usr/local/bin, or in ~/.local/bin if that one is not writable. Windows already open will not see it.

3. Check it works

   Run verbiqo --version.

Everyday examples

Modes for --mode: grammar, professional, friendly, concise, technical, sanitize, summarize and key-points.

Which AI service it uses

The command uses the AI service you set up in the app’s Settings, including any endpoint you confirmed there. To use a different one for a single command, add --provider, --endpoint and --model; these override the saved settings but never change them. Connected systems are the ones you set up in the app, or with the commands below.

It takes the rest from the app too, so nothing goes in a shell profile:

• The API key saved in Settings. On a Mac, the first time, macOS asks whether the terminal may read it from the Keychain: choose Always Allow.
• Certificates: the app’s Trusted CA and everything in the computer’s trust store, so a gateway behind your organization’s certificate works as it does in the app.
• Prometheus’s semantic index: the embedding endpoint, model and key from Extensions › Prometheus.

Any of them can be set for one shell instead: variables for a terminal.

Chat with your systems

verbiqo chat attaches a connection with a flag, one per system, by the name you gave it: --pg, --mysql, --mssql, --redis, --es, --k8s, --prom, --splunk, and --security all (or one system) for the security review. Up to three at once.

verbiqo chat --mysql shop --prom prod "why was the shop slow last night"
cat deploy.log | verbiqo chat --k8s prod "did this deploy cause the restarts"

Inside the conversation, a line starting with / is a command; anything else is a question.

When a name you used matches several things, the terminal numbers the choices and says Type a number; nothing is read until you do. More

Your systems

Every system has list, add, test and remove, a run for one of its named reports, and a way to send one statement, command or request yourself. Secrets are piped in with --password-stdin, --token-stdin or --secret-stdin, or read from a variable with the -env form, never typed on the command line; they go to the Keychain or Credential Manager. verbiqo pg help, and the same for each system, prints every flag and the reports run takes.

verbiqo pg     list | add | remove | test | run <name> <diagnostic> | query <name> "SELECT ..."
verbiqo mysql  list | add | remove | test | run | query          (--socket for a login on the database node)
verbiqo mssql  list | add | remove | test | run | query          (--instance NAME for a named instance)
verbiqo redis  list | add | remove | test | run | command <name> "INFO memory"
verbiqo es     list | add | remove | test | review | analyze | run | request <name> "GET _cat/shards"
verbiqo k8s    list | contexts | add | remove | test | run | command <name> "get pods -n shop" | readiness
verbiqo prom   list | add | remove | test | describe | run | query | range | ask <name> "why was mysql down yesterday"
               index-build | index-refresh | index-status | index-discover | index-inspect | index-prefixes | index-prefix-refresh | index-clear
               cache-path | cache-clear
verbiqo splunk list | add | login | remove | test | indexes | shapes | ask <name> "..." | locate <name> <cluster>

• es review runs every rule of the cluster review; es analyze reads a search from a file or stdin and rewrites it.
• k8s contexts lists the contexts in a kubeconfig; k8s readiness prints the read-only ClusterRole, service account and binding to create.
• prom index-build builds the local map a Prometheus connection needs before its first question (The local map).
• splunk login exchanges a directory password for a session key once; the password is never stored.

Each page under Set up has the add line for its system, TLS flags included: Encrypted connections.

Security review and access record

verbiqo security review [--system pg|mysql|mssql|redis|es|k8s|prom|splunk] [--connection NAME]
                        [--format text|json|csv|html] [--output FILE] [--all] [--fail-on-findings]
verbiqo audit show [--since 24h|7d|DATE] [--system ...] [--origin model|verbiqo|person] [--connection NAME] [--limit N] [--json]
verbiqo audit export --format csv|json [--since ...] [--output FILE]
verbiqo audit verify | forward | path

security review checks every saved connection, or those named, against each control and reports pass, fail or not checked, mapped to CIS Controls v8, ISO 27001:2022 and SOC 2; --fail-on-findings makes it exit non-zero for a pipeline. Security review. audit reads the record of every statement, command and request Verbiqo sent to your systems, checks it has not been altered, sends what is pending to your SIEM, or prints where it is kept. Access record.

Check the model

Whether the model in Settings, or another you name, is fit for Verbiqo: it runs Verbiqo’s own steps on built-in data and reports pass or fail, with timings and what to change. It calls only the model.

verbiqo model-check
verbiqo model-check --provider openai --endpoint http://gpu-box:8000/v1 --model "Qwen/Qwen3-32B-AWQ"

See Choose and check a model for what is checked and the models tested so far.

Make an organization policy

For administrators, with no licence needed: start from a template, sign it with your organization’s key, and check what a signed policy allows.

verbiqo policy keygen policy-private.pem policy-public.pem
verbiqo policy template > policy.json
verbiqo policy sign policy.json policy-private.pem policy-signed.json
verbiqo policy check policy-signed.json policy-public.pem

See Privacy policy. To make the files that set and lock settings on Macs and Windows PCs from one list of values: verbiqo policy settings template and verbiqo policy settings build (Make the files).

Manage a licence’s seats

For whoever holds a licence key file, with no licence needed on the computer it runs on:

verbiqo licence seats team.verbiqo-license
verbiqo licence free team.verbiqo-license DEVICE-ID

The list shows each computer’s name on the licence (set in its Settings › Licence, its computer name unless changed), its device ID, when it was activated, when it last checked in (to the minute, UTC), and from when another computer can free its seat.

On Linux servers

Team and organization licences only.

The command line tool also runs on Linux servers, for asking questions where your logs and services live. It comes as a tarball per architecture (x64 and arm64) with everything it needs inside, so nothing is installed on the host and the JavaScript stays readable beside it. Unpack it and run ./install.sh /usr/local/bin, or use it from the folder as ./verbiqo. It works only with a team or organization licence and has no trial: put the licence file at /etc/verbiqo/license.json for a fleet or ~/.config/verbiqo/license.json for one user, or name it in VERBIQO_LICENSE_PATH. Connection details are kept in a file only your user can read, so use read-only credentials (Where sign-in details are kept). To get it, mention it when you request a team or organization licence.

All options

Run verbiqo --help for the commands, the chat flags and the rewrite options. Each system prints its own verbs, flags and reports with verbiqo pg help (and mysql, mssql, redis, es, k8s, prom, splunk); verbiqo security and verbiqo audit with nothing after them print theirs, as do verbiqo policy and verbiqo policy settings for the administrator commands. Inside verbiqo chat, type /help.

Like the app, the command keeps no history of what you asked or received, and sends your text only to the AI service you chose. What it read from your systems is recorded in the access record, never the question or the answer.