WebSocket and streaming

The WebSocket server streams analysis results, trace events and state changes to remote clients in real time using a binary FlatBuffers protocol.

Start the server

./tenet trace.bin --ws-port 9090

Protocol

• Transport: WebSocket on the configured port.
• Schema: FlatBuffers (.fbs definitions in the Tenet source). Binary encoding, no JSON parsing overhead.
• Direction: Server pushes events; clients subscribe to topics.

Event topics

Topic	Content
timeline	Instruction cursor movement, navigation events
pass_state	Pass state changes (started, progress, completed, failed)
analysis_result	Structured pass results as they complete
register_update	Register value changes at the current cursor
memory_update	Memory region changes
breakpoint	Breakpoint/watchpoint hit events

Usage patterns

• Live dashboard: Build a web-based trace monitor that updates as analysis proceeds.
• Remote collaboration: Multiple clients connect to the same Tenet instance; all see the same cursor and results.
• CI integration: A CI job opens a trace, runs passes, streams results and exits.
• Custom visualization: Render CFG, taint graphs or memory heatmaps in a web UI using the streamed events.

Connection lifecycle

1. Client connects to ws://host:port.
2. Client subscribes to one or more topics.
3. Server pushes events for subscribed topics.
4. Client can also send JSON-RPC requests on the same connection for interactive queries.
5. On disconnect, the server continues the session; the next client reconnects to the same state.

Limitations

• WebSocket streaming requires an active Tenet process; it is not an offline API.
• FlatBuffers schema stability is not guaranteed across versions; clients may need updates when upgrading Tenet.
• High-frequency events (e.g., timeline cursor at every instruction) may need client-side throttling or debouncing.