feat: preempt active runs in interrupt queue mode

docs/api/PROTOCOL.md

@@ -34,7 +34,7 @@ The gateway serialises agent work **per session**, not per WebSocket connection:
 
 - Requests that target the same `sessionId` run one-at-a-time (FIFO) in a per-session lane.
 - Requests for different sessions can run in parallel.
-- Lane policy is configurable (`collect`, `followup`, `steer_backlog`, `interrupt`) with per-channel and per-session overrides.
+- Lane policy is configurable (`collect`, `followup`, `steer`, `steer_backlog`, `interrupt`) with per-channel and per-session overrides.
 - Session-local overrides can be managed at runtime via `agent.send` commands: `/queue`, `/queue set ...`, `/queue reset`.
 
 This is implemented via a per-lane queue (`LaneQueue`) in the gateway server, and used by `agent.send` and `agent.cancel`.
@@ -69,6 +69,8 @@ sequenceDiagram
   G-->>C: result.cancelled=true/false
 ```
 
+`interrupt` queue mode also requests active-run cancellation when a newer request is enqueued for the same session lane. Cancellation still completes at agent/tool-loop safe points.
+
 ### Base URL
 
 - WebSocket: `ws://localhost:18800` (or `wss://` if using TLS)

docs/plans/state.json

@@ -5163,10 +5163,26 @@
         "docs/plans/state.json"
       ],
       "test_status": "Docs-only change (no code paths affected)"
+    },
+    "queue-interrupt-active-cancel": {
+      "status": "completed",
+      "date": "2026-02-18",
+      "updated": "2026-02-18",
+      "summary": "Implemented Tier A queue-interrupt semantics hardening in gateway routing: interrupt mode now requests cancellation of active session work when a newer request arrives, while preserving best-effort safe-point cancellation. Added session-level cancellation API, handler integration, regression tests, and docs updates.",
+      "files_modified": [
+        "src/gateway/session-bridge.ts",
+        "src/gateway/session-bridge.test.ts",
+        "src/gateway/handlers/agent.ts",
+        "src/gateway/handlers/agent.test.ts",
+        "README.md",
+        "docs/api/PROTOCOL.md",
+        "docs/plans/state.json"
+      ],
+      "test_status": "pnpm test:run src/gateway/session-bridge.test.ts src/gateway/handlers/agent.test.ts + pnpm typecheck passing"
     }
   },
   "overall_progress": {
-    "total_test_count": 1900,
+    "total_test_count": 1903,
     "all_tests_passing": true,
     "p0_completion": "3/3 (100%)",
     "p1_completion": "4/4 (100%)",
@@ -5186,7 +5202,7 @@
     "gmail_auth_cli": "flynn gmail-auth command implemented with OAuth2 flow, doctor check, config routed to Telegram",
     "native_audio_support": "completed — smart routing for native audio (Gemini/OpenAI/GitHub) vs Whisper transcription fallback",
     "remaining_phases_completion": "Phase 1: 3/3 (100%) — context levels, command registry, memory structure. Phase 2: 3/3 (100%) — component registry, confidence routing, history index. Phase 3: 2/2 (100%) — adaptive memory/compaction, truthfulness/autonomy hardening",
-    "next_up": "Monitor production feedback for bidirectional session-transfer command behavior across Telegram/TUI and prioritize the next post-parity reliability/capability slice"
+    "next_up": "Implement Tier A2 from the OpenClaw roadmap: daily memory-log cadence + proactive extraction beyond compaction-only paths"
   },
   "soul_md_and_cron_create": {
     "date": "2026-02-11",