Keyboard Shortcuts
Keyboard shortcuts are available throughout the application. Press ? (when not typing in an input field) to open the shortcuts reference at any time.| Action | macOS | Windows / Linux |
|---|---|---|
| New analysis | Cmd + N | Ctrl + N |
| Toggle sidebar | Cmd + Shift + S | Ctrl + Shift + S |
| Toggle Ask/Agent mode | Cmd + / | Ctrl + / |
| Close modal or panel | Escape | Escape |
| Open shortcuts reference | ? | ? |
Shortcuts are suppressed when focus is inside a text input, textarea, or any editable field — so they won’t interfere with typing.
Conversation Branching
You can fork a conversation from any message to explore alternative analysis paths without losing your original thread. How it works:- Hover over any message in the chat to reveal the branch button
- Click it to create a new conversation branch starting from that point
- The branch inherits the full context up to that message — triage data, tool outputs, and prior findings
- The original conversation remains untouched
- You want to try a different analysis approach without losing the current one
- The agent went down a path you want to revisit with a different question
- You want to ask follow-up questions about an intermediate finding while the main analysis continues
Chat Sharing
Share an analysis session with colleagues by generating a share link with a configurable expiry. How it works:- Click the share button in the chat header
- Choose an expiry duration for the link
- Copy the generated URL and send it to your team
Ghidra Project Persistence
When analysing native binaries in Agent mode, the Ghidra project is preserved across messages within the same chat session. This means:- The initial Ghidra import and analysis (which can take minutes for large binaries) only happens once
- Follow-up questions like “decompile the main function” or “show cross-references to CreateRemoteThread” reuse the already-analysed project instantly
- Switching to a new chat session creates a fresh project
Auto-Continue
When an Agent mode analysis requires more steps than fit in a single response, the agent can automatically continue its work. You’ll see an auto-continue banner indicating the agent is still working. This is useful for complex analyses where the agent needs to:- Chain many tools together (e.g., decompile → trace calls → extract IOCs → generate YARA rule)
- Process large outputs that require multiple follow-up tool calls
- Iterate on findings to build a complete picture before reporting back