fix(agent): recover when action intent has no tool call

docs/plans/state.json

@@ -3,6 +3,18 @@
   "updated_at": "2026-02-23",
   "description": "Tracks the status of all Flynn plans and implementation phases",
   "plans": {
+    "toolloop-action-intent-recovery": {
+      "status": "completed",
+      "date": "2026-02-23",
+      "updated": "2026-02-23",
+      "summary": "Hardened NativeAgent tool-loop behavior when models narrate imminent actions without emitting tool calls. Added a one-time in-loop nudge for action-intent text with missing tool calls, so Flynn re-prompts the model to either execute the tool now or return an explicit blocker in the same turn.",
+      "files_modified": [
+        "src/backends/native/agent.ts",
+        "src/backends/native/agent.test.ts",
+        "docs/plans/state.json"
+      ],
+      "test_status": "pnpm test:run src/backends/native/agent.test.ts + pnpm typecheck passing"
+    },
     "gmail-filter-full-scope-auth": {
       "status": "completed",
       "date": "2026-02-23",
@@ -6287,7 +6299,7 @@
     }
   },
   "overall_progress": {
-    "total_test_count": 1958,
+    "total_test_count": 1960,
     "all_tests_passing": true,
     "p0_completion": "3/3 (100%)",
     "p1_completion": "4/4 (100%)",
@@ -6307,6 +6319,7 @@
     "dashboard_observability": "completed — service health graphs + core service log viewer added to web UI via observability RPCs and bounded backend sampling",
     "gmail_auth_cli": "flynn gmail-auth command implemented with OAuth2 flow, doctor check, config routed to Telegram",
     "gmail_filter_creation": "completed — gmail.filter.create tool added with criteria/action validation; gmail-auth now requests full Gmail scope (https://mail.google.com/) for complete filter permissions",
+    "toolloop_action_intent_recovery": "completed — when a model claims it will execute a tool but emits no tool call, NativeAgent now issues one internal nudge and continues the same turn to execute tools or produce a concrete blocker",
     "native_audio_support": "completed — smart routing for native audio (Gemini/OpenAI/GitHub) vs Whisper transcription fallback, plus 2026-02-23 arg hydration hardening, tool.args_rewritten audit metric, transient fetch retry/timeout hardening, localhost->127.0.0.1 fallback for transcription endpoint connectivity, and whisper docker-compose entrypoint arg fix for port 18801",
     "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": "Track OpenClaw evolution regularly for inspiration and feature ideas"