`dxb ai ask` brings the [site assistant](/reader/site-assistant) to your terminal. Ask a question in plain language and get an answer grounded in your documentat...
`dxb ai ask` brings the [site assistant](/reader/site-assistant) to your terminal. Ask a question in plain language and get an answer grounded in your documentat...
dxb ai ask brings the site assistant to your terminal. Ask a question in plain language and get an answer grounded in your documentation, with citations — without leaving your shell. It's useful for quick lookups, scripting doc-aware tooling, and letting AI coding agents query your docs.
Ask a question
dxb ai ask "How do I rotate API keys?" -p my-docsEvery answer is grounded in your published documentation and cites the pages it drew from.
Options
| Flag | Description |
|---|---|
| `--locale <locale>` | Answer in a specific locale. |
| `-p`, `--project <slug>` | Which project's docs to answer from. |
| `--json` | Structured output (answer + citations) for scripting (global flag). |
# Answer in French
dxb ai ask "Comment publier une page ?" --locale fr -p my-docs
# JSON for scripts
dxb ai ask "How do I publish?" --json | jq -r '.answer'Grounding
dxb ai ask answers only from your documentation. If your docs don't cover something, it says so rather than guessing — the same grounding behavior as the site assistant.
ask vs. search
| Col 1 | `dxb ai ask` | [`dxb search`](/cli/inspect/search-and-quality) |
|---|---|---|
| Input | A natural-language question | Keywords |
| Output | A synthesized, cited answer | A ranked list of pages |
| Best for | "How do I…?" | "Find pages about X" |
For AI agents
Because it's scriptable, dxb ai ask is a building block for doc-aware tooling. AI coding agents can consult your documentation while working — though MCP-native clients usually call the MCP tools directly. See the skill + MCP setup.