Tool approval panel

Agent tool-call approval cards with explicit decisions, hold-to-confirm destructive actions, execution states, denial and retry controls.

Approval gate for a proposed assistant bulk change showing sample before and after field values with approve and reject buttons.

Vertical timeline of agent tool steps with status icons, tool badges, mono query detail lines, durations, and a raw log link.

AI token usage panel with a monthly progress bar, a per feature breakdown, and the overage policy stated in the same view.

Conversation thread with alternating user and assistant bubbles, inline source chips, and regenerate and copy affordances.

Pre-conversation assistant panel with 4 example prompt chips and plain lists of what the assistant can see and cannot do yet.

Sources side panel listing 4 cited documents with title, matched snippet, relevance chip, and an honest footer note.

Data access panel listing assistant readable sources with icon, item count, last sync time, and a switch, plus one excluded source.

Chat history sidebar listing past assistant threads grouped by recency with title, snippet, timestamp, and a delete affordance.

AI response feedback row with thumbs up and down, a copy and regenerate action, and a reason picker revealed on negative feedback.

Ruled rows pairing each safety rule with an icon, explanation, and switch, above a blocked topics list and an audit logging note.

Text editor surface showing an AI ghost-text completion with Tab to accept and Esc to dismiss keyboard hints and an authorship notice.