Hibryda
3f1638c98b
fix: resolve medium/low audit findings across backend and frontend
...
- ctx CLI: validate int() limit arg, wrap FTS5 MATCH in try/except
- ctx.rs: FTS5 error message clarity, Mutex::lock() returns Err not panic
- sdk-messages.ts: runtime type guards (str/num) replace bare `as` casts
- agent-runner.ts: strip ANTHROPIC_* env vars alongside CLAUDE*
- agent-dispatcher.ts: timestamps use seconds (match session.rs convention)
- remote.rs: disconnect handler uses lock().await not try_lock()
- session.rs: propagate pane_ids serialization error
- watcher.rs: reject root-level paths instead of silent no-op
- lib.rs: log warnings on profile.toml read failure and resource_dir error
- agent-bridge.ts: validate event payload is object before cast
2026-03-08 20:10:54 +01:00
Hibryda
4bdb74721d
fix(security): audit fixes — path traversal, race conditions, memory leaks, transaction safety
...
- lib.rs: claude_read_skill path traversal prevention (canonicalize + starts_with)
- agent-dispatcher.ts: re-entrancy guard on exit handler, clear maps in stop
- machines.svelte.ts: track UnlistenFn array + destroyMachineListeners()
- agent-runner.ts: controller.signal.aborted, async handleMessage + .catch()
- remote.rs: try_lock → async lock, abort tasks on remove
- session.rs: unchecked_transaction for save_agent_messages
- agent-bridge.ts: safe msg.event check (implicit in dispatcher changes)
2026-03-08 20:03:50 +01:00
Hibryda
71100da125
feat(v2): refactor reconnection probe to TCP-only and add frontend listeners
...
Replace attempt_ws_connect() with attempt_tcp_probe() in RemoteManager to
avoid allocating per-connection resources (PtyManager, SidecarManager) on
the relay during reconnection probes. Add onRemoteMachineReconnecting and
onRemoteMachineReconnectReady event listeners in remote-bridge.ts. Wire
machines store to auto-reconnect when relay becomes reachable.
2026-03-06 21:50:45 +01:00
Hibryda
b0cce7ae4f
feat(v2): add relay response propagation and reconnection with exponential backoff
...
Relay (bterminal-relay): command handlers now send structured responses
(pty_created, pong, error) back via shared event channel with commandId
for correlation. New send_error() helper replaces log-only error
reporting.
RemoteManager (remote.rs): exponential backoff reconnection on
disconnect (1s/2s/4s/8s/16s/30s cap). Uses attempt_ws_connect() probe
with 5s timeout. Emits remote-machine-reconnecting and
remote-machine-reconnect-ready events. Handles pty_created relay event
as remote-pty-created Tauri event.
2026-03-06 19:49:19 +01:00
Hibryda
0b39133d66
feat(v2): add RemoteManager for multi-machine WebSocket connections
...
New remote.rs module in src-tauri with WebSocket client connections to
bterminal-relay instances. Machine lifecycle: add/remove/connect/
disconnect. 12 new Tauri commands for remote operations. Heartbeat
ping every 15s. lib.rs updated with remote commands and AppState.
2026-03-06 19:05:47 +01:00
