fix: avoid stuck probe failures for unavailable activity (#129)

.changeset/fix-activity-unavailable-stuck.md

@@ -0,0 +1,5 @@
+---
+"@aoagents/ao-core": patch
+---
+
+Stop carrying forward `stuck` / `probe_failure` session truth when the runtime is still confirmed alive and activity is merely unavailable, and degrade that combination to `detecting` until stronger evidence arrives.

packages/core/src/__tests__/lifecycle-manager.test.ts

@@ -542,6 +542,27 @@ describe("check (single session)", () => {
     expect(meta?.["lifecycleEvidence"]).toContain("activity_signal=probe_failure");
   });
 
+  it("degrades stuck probe-failure sessions to detecting when runtime is alive but activity is unavailable", async () => {
+    const registryWithoutAgent: PluginRegistry = {
+      ...mockRegistry,
+      get: vi.fn().mockImplementation((slot: string) => {
+        if (slot === "runtime") return plugins.runtime;
+        return null;
+      }),
+    };
+
+    const lm = setupCheck("app-1", {
+      session: makeSession({ status: "stuck" }),
+      registry: registryWithoutAgent,
+    });
+
+    await lm.check("app-1");
+
+    expect(lm.getStates().get("app-1")).toBe("detecting");
+    const meta = readMetadataRaw(env.sessionsDir, "app-1");
+    expect(meta?.["lifecycleEvidence"]).toContain("activity_signal=unavailable");
+  });
+
   it("detects needs_input from agent", async () => {
     vi.mocked(plugins.agent.getActivityState).mockResolvedValue({ state: "waiting_input" });
 

packages/core/src/lifecycle-manager.ts

@@ -909,6 +909,17 @@ export function createLifecycleManager(deps: LifecycleManagerDeps): LifecycleMan
         lifecycle.session.state === "stuck" ||
         lifecycle.session.state === "needs_input")
     ) {
+      const preservingProbeFailureStuck =
+        activitySignal.state === "unavailable" &&
+        lifecycle.session.state === "stuck" &&
+        lifecycle.session.reason === "probe_failure" &&
+        lifecycle.runtime.state === "alive";
+
+      if (preservingProbeFailureStuck) {
+        setSessionState("detecting", "probe_failure");
+        return commit(SESSION_STATUS.DETECTING, activityEvidence, 0);
+      }
+
       return commit(deriveLegacyStatus(lifecycle, session.status), activityEvidence, 0);
     }