Skip to content

Working in the Sidekick View

The Sidekick view is where you work with Sidekick. It is the home of the lead agent — your single point of contact for the entire workspace. Ask it anything about the binary and it answers directly, delegating to specialist agents behind the scenes when the work benefits. You do not start or dig through separate chats, and you do not manage a context window: one persistent conversation carries your analysis forward, with the session briefing, findings, and recorded outcomes in adjacent tabs.

This is the intended home base for AI-first reverse engineering. When you open a binary, talk to Sidekick here — the view opens as the center-pane view by default so it is the first thing you see.

The Sidekick view in the center pane with the Chat tab active, showing a two-message conversation with expandable tool groups, findings, and code references

The Sidekick view and the Chats sidebar

Sidekick gives you two ways to converse, and they serve different purposes:

  • The Sidekick view (this page) is the center-pane home of the lead agent. It holds one persistent Workspace Chat — a single ongoing conversation for the whole workspace — plus the briefing, artifacts, and findings. This is where most users work day to day. You never have to decide which chat to open or worry about losing context between them.
  • The Chats sidebar is the traditional, thread-centric surface: a full list of conversations you can start, revisit, and branch yourself. It is also where the specialist agents the lead agent spawns appear, each as its own thread.

Reach for the Chats sidebar when you want flexibility — to run and manage several independent conversations by hand — or transparency — to watch or steer exactly what the delegated agents are doing. For everyday analysis, stay in the Sidekick view and let the lead agent coordinate the rest.

The Sidekick view is a center-pane view, alongside Binary Ninja's Linear, Graph, Hex, and IL views. Everything it shows is also reachable from the individual Sidekick sidebars; the difference is emphasis — the view puts the lead agent front and center, while the sidebars let you dock a single surface alongside a code view.

Opening the Sidekick View

You can open the Sidekick view in any of these ways:

  • Choose Plugins > Sidekick > Open Sidekick View. This opens the view on its Chat tab, or focuses it and switches to the Chat tab if it is already open.
  • Choose Plugins > Sidekick > Open Session Briefing. This opens the same Sidekick view, but switches directly to its Briefing tab.
  • Select Sidekick from the view-type selector at the top of the view pane, alongside options like Linear and Graph.

Note

The view needs an active Sidekick session for the binary before it can show content. If you open it immediately after Binary Ninja starts, or before the session finishes connecting, you may briefly see a "Sidekick is starting..." placeholder. Each tab also loads its content the first time you switch to it, so a newly opened tab can show a brief "Loading..." placeholder.

Setting It as Your Default View

By default, Sidekick opens as the default view the first time you open a binary — before you have switched to any other view for that file. This is controlled by the sidekick.ui.prefer_sidekick_view setting, which is enabled by default.

This behavior applies once per open file: after Sidekick has shown its view for a file, later view-type changes for that file (switching tabs, splitting panes, navigating between synced views) no longer favor Sidekick. Closing and reopening the file resets it, so the next time you open that file, the Sidekick view is offered as the default again.

To turn this off, open Edit > Preferences > Settings, search for "Sidekick", and disable Prefer Sidekick View. With the setting disabled, Binary Ninja opens binaries in its normal default view, and you can still switch to the Sidekick view manually at any time using the methods above. See Settings Reference for the full setting description.

Chat Tab

The Chat tab holds a single, persistent conversation for the current workspace — a thread named Workspace Chat — rather than the full thread list you see in the Chat sidebar. Until you send your first message in it, the tab shows a home state instead of an empty transcript: a summary of the binary you're analyzing (architecture, platform, size, entry point) and a handful of suggested questions you can click to send immediately.

Above the conversation, a collapsible work queue drawer appears whenever there is active work to track. Each row is either:

  • An active notebook entry with open tasks, showing a progress bar for the percentage of tasks completed on that entry and how long its most recent chat thread has been working
  • An open blocker, shown with a "Blocker" chip and how long ago it was raised

The drawer is hidden entirely when there is no active work or open blockers. Clicking a task row opens the chat thread currently working on it — in this tab if it's the Workspace Chat thread, or in the Chat sidebar otherwise. Clicking a blocker row opens its notebook entry in Resources, focused on that blocker.

For everything else about chatting with Sidekick — the message composer, @-mentions, tool approvals, and the full multi-thread list — see Analyzing Binaries with Chat.

The Chat tab with the work queue drawer expanded at the top, showing an open-blocker row with a "Blocker" chip ("Confirm which host-key slot is used when multiple are present") above a task row with a 50% progress bar for the SSH algorithm-inventory entry, with the conversation below

Briefing Tab

The Briefing tab shows the Session Briefing for the current workspace, with a compact binary-context summary above it. The briefing renders as a read-only document — you cannot type edits directly into it. Sidekick keeps it current automatically as your notebook activity accumulates, and you can ask for specific changes in chat if something needs correcting.

See Session Briefing for what each of the briefing's six sections contains.

The Briefing tab with the binary-context summary above the rendered session briefing, showing the Goal, Status, and Key Findings sections

Artifacts Tab

The Artifacts tab is a feed of notebook artifact outcomes — concrete deliverables like YARA signatures, reports, and generated scripts — drawn from every entry in the workspace. Only draft and verified outcomes appear; rejected outcomes are excluded from the feed.

Each card is the same outcome card you see in the notebook detail view, including its status rail, provenance edge lines, references, and Ask in chat button. See Working with Outcomes for how to read a card and how outcomes move through their status lifecycle.

The Artifacts tab with a YARA-rule artifact expanded to show its content, verified status, provenance, and references, above a collapsed report card

Findings Tab

The Findings tab works the same way as the Artifacts tab, but shows notebook finding outcomes instead — reusable, anchored facts discovered during analysis. Only draft and verified findings appear; rejected findings are excluded.

The Findings tab showing two finding cards, one expanded to reveal its full anchored description with code references