Watching your agent work

While you drive your AI tool from the terminal, three tabs turn its work into something you can actually read. Activity is the story of what the agent did, prompt by prompt. Timeline is the raw event stream you can search and filter. Tasks is the checklist the agent is working through, built for you automatically. Together they mean you never have to scroll back through a wall of terminal output to understand what happened.

Activity — traces

The Activity tab is the heart of Operon. Every prompt you send becomes a Trace: a single card that groups everything the AI did in response — the tools it ran, the files it edited, the sub-agents it spawned, any errors it hit, the cost it incurred, and its final written response. Instead of a firehose of scrollback, you get one card per turn that you can expand to see exactly what unfolded.

• Tool timeline — expand a trace to see each tool execution in order, with smart summaries: the command that ran, how many lines changed, the search pattern used, or the task list written.
• File diffs — edits render as an inline diff viewer with syntax highlighting; expand any diff to a full-screen modal to read it in detail.
• Sub-agents — when the agent delegates to a sub-agent, that appears as its own block inside the trace with its status and progress.
• Cost — each trace shows the estimated tokens and spend for that turn, so you can spot the expensive prompts.
• The response — the AI’s final answer is rendered as formatted markdown at the end of the trace.

A Session Intelligence sidebar runs alongside the trace stream, surfacing the facts and decisions Operon extracted from the conversation. A density toggle switches between a comfortable and a compact layout depending on how much you want on screen at once.

1. 1

   Open the Activity tab

   From an open session, click Activity (or press Cmd+3). You see the trace stream, newest activity at the top.

2. 2

   Expand a trace

   Click a trace card to expand it. The tool execution timeline unfolds, showing each step the agent took in order with a smart summary for each.

3. 3

   Read the diffs

   For any file edit, the inline diff shows exactly what changed. Click to expand it to a full-screen modal when you want a closer look.

4. 4

   Inspect the details

   Click a trace to open the properties panel on the right for full inspection — the complete tool inputs, outputs, cost breakdown, facts, and decisions for that turn.

Timeline — the event stream

Where Activity groups everything into per-prompt traces, the Timeline gives you the flat, chronological stream of every individual event: prompts, tool calls, file changes, git activity, errors, sub-agent runs, context snapshots, and session milestones. It is grouped by trace so related events stay together, but it is built for scanning and searching across the whole session.

• Search — type to filter the stream to just the events that match.
• Category filter pills — narrow to a single category such as tools, errors, files, git, agents, or context with one click.
• Sort toggle — flip between newest-first and oldest-first order.

Tasks — the kanban

The Tasks tab is a four-column kanban board that tracks the work in a session: To Do, In Progress, Blocked, and Done. You do not have to build it. Operon populates it two ways — by extracting tasks from the conversation, and by syncing directly from the agent’s own todo list as it writes one.

Every task card has a send-to-terminal action: click it to drop that task back into the running terminal as a prompt for the agent. It is a quick way to hand the next item to your tool without retyping it. The board is available as a kanban or a plain list, whichever you prefer.

Related