From c90cb9c3c95cc2f8e86efead0c9ea05565a85c53 Mon Sep 17 00:00:00 2001
From: Peter Steinberger <steipete@gmail.com>
Date: Thu, 9 Apr 2026 05:26:07 +0100
Subject: [PATCH] test: expand helper coverage for memory flush policy

---
 .../reply/agent-runner-memory.test.ts         | 92 +++++++++++++++++++
 src/auto-reply/reply/agent-runner-memory.ts   |  4 +-
 2 files changed, 95 insertions(+), 1 deletion(-)

diff --git a/src/auto-reply/reply/agent-runner-memory.test.ts b/src/auto-reply/reply/agent-runner-memory.test.ts
index be17c62d528..8196df99d2b 100644
--- a/src/auto-reply/reply/agent-runner-memory.test.ts
+++ b/src/auto-reply/reply/agent-runner-memory.test.ts
@@ -195,4 +195,96 @@ describe("runMemoryFlushIfNeeded", () => {
     expect(persisted.main.memoryFlushCompactionCount).toBe(2);
     expect(persisted.main.memoryFlushAt).toBe(1_700_000_000_000);
   });
+
+  it("skips memory flush for CLI providers", async () => {
+    const sessionEntry: SessionEntry = {
+      sessionId: "session",
+      updatedAt: Date.now(),
+      totalTokens: 80_000,
+      compactionCount: 1,
+    };
+
+    const entry = await runMemoryFlushIfNeeded({
+      cfg: { agents: { defaults: { cliBackends: { "codex-cli": { command: "codex" } } } } },
+      followupRun: createFollowupRun({ provider: "codex-cli" }),
+      sessionCtx: { Provider: "whatsapp" } as unknown as TemplateContext,
+      defaultModel: "codex-cli/gpt-5.4",
+      agentCfgContextTokens: 100_000,
+      resolvedVerboseLevel: "off",
+      sessionEntry,
+      sessionStore: { main: sessionEntry },
+      sessionKey: "main",
+      isHeartbeat: false,
+      replyOperation: createReplyOperation(),
+    });
+
+    expect(entry).toBe(sessionEntry);
+    expect(runEmbeddedPiAgentMock).not.toHaveBeenCalled();
+  });
+
+  it("uses configured prompts and stored bootstrap warning signatures", async () => {
+    const sessionEntry: SessionEntry = {
+      sessionId: "session",
+      updatedAt: Date.now(),
+      totalTokens: 80_000,
+      compactionCount: 1,
+      systemPromptReport: {
+        source: "run",
+        generatedAt: Date.now(),
+        systemPrompt: { chars: 1, projectContextChars: 0, nonProjectContextChars: 1 },
+        injectedWorkspaceFiles: [],
+        skills: { promptChars: 0, entries: [] },
+        tools: { listChars: 0, schemaChars: 0, entries: [] },
+        bootstrapTruncation: {
+          warningMode: "once",
+          warningShown: true,
+          promptWarningSignature: "sig-b",
+          warningSignaturesSeen: ["sig-a", "sig-b"],
+          truncatedFiles: 1,
+          nearLimitFiles: 0,
+          totalNearLimit: false,
+        },
+      },
+    };
+    registerMemoryFlushPlanResolver(() => ({
+      softThresholdTokens: 4_000,
+      forceFlushTranscriptBytes: 1_000_000_000,
+      reserveTokensFloor: 20_000,
+      prompt: "Write notes.\nNO_REPLY to memory/2023-11-14.md and MEMORY.md",
+      systemPrompt: "Flush memory now. NO_REPLY memory/YYYY-MM-DD.md MEMORY.md",
+      relativePath: "memory/2023-11-14.md",
+    }));
+
+    await runMemoryFlushIfNeeded({
+      cfg: { agents: { defaults: { compaction: { memoryFlush: {} } } } },
+      followupRun: createFollowupRun({ extraSystemPrompt: "extra system" }),
+      sessionCtx: { Provider: "whatsapp" } as unknown as TemplateContext,
+      defaultModel: "anthropic/claude-opus-4-6",
+      agentCfgContextTokens: 100_000,
+      resolvedVerboseLevel: "off",
+      sessionEntry,
+      sessionStore: { main: sessionEntry },
+      sessionKey: "main",
+      isHeartbeat: false,
+      replyOperation: createReplyOperation(),
+    });
+
+    const flushCall = runEmbeddedPiAgentMock.mock.calls[0]?.[0] as {
+      prompt?: string;
+      extraSystemPrompt?: string;
+      bootstrapPromptWarningSignaturesSeen?: string[];
+      bootstrapPromptWarningSignature?: string;
+      memoryFlushWritePath?: string;
+      silentExpected?: boolean;
+    };
+    expect(flushCall.prompt).toContain("Write notes.");
+    expect(flushCall.prompt).toContain("NO_REPLY");
+    expect(flushCall.prompt).toContain("MEMORY.md");
+    expect(flushCall.extraSystemPrompt).toContain("extra system");
+    expect(flushCall.extraSystemPrompt).toContain("Flush memory now.");
+    expect(flushCall.memoryFlushWritePath).toBe("memory/2023-11-14.md");
+    expect(flushCall.silentExpected).toBe(true);
+    expect(flushCall.bootstrapPromptWarningSignaturesSeen).toEqual(["sig-a", "sig-b"]);
+    expect(flushCall.bootstrapPromptWarningSignature).toBe("sig-b");
+  });
 });
diff --git a/src/auto-reply/reply/agent-runner-memory.ts b/src/auto-reply/reply/agent-runner-memory.ts
index ce600e3417e..e45c45e57a3 100644
--- a/src/auto-reply/reply/agent-runner-memory.ts
+++ b/src/auto-reply/reply/agent-runner-memory.ts
@@ -46,6 +46,7 @@ import type { ReplyOperation } from "./reply-run-registry.js";
 import { incrementCompactionCount } from "./session-updates.js";
 
 const memoryDeps = {
+  compactEmbeddedPiSession,
   runWithModelFallback,
   runEmbeddedPiAgent,
   registerAgentRunContext,
@@ -59,6 +60,7 @@ const memoryDeps = {
 export function setAgentRunnerMemoryTestDeps(overrides?: Partial<typeof memoryDeps>): void {
   Object.assign(memoryDeps, {
     runWithModelFallback,
+    compactEmbeddedPiSession,
     runEmbeddedPiAgent,
     registerAgentRunContext,
     refreshQueuedFollowupSession,
@@ -432,7 +434,7 @@ export async function runPreflightCompactionIfNeeded(params: {
     params.sessionKey ?? params.followupRun.run.sessionKey,
     { storePath: params.storePath },
   );
-  const result = await compactEmbeddedPiSession({
+  const result = await memoryDeps.compactEmbeddedPiSession({
     sessionId: entry.sessionId,
     sessionKey: params.sessionKey,
     allowGatewaySubagentBinding: true,