diff --git a/CHANGELOG.md b/CHANGELOG.md
index 2d0a7022c28..eac5c3d2463 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -18,6 +18,7 @@ Docs: https://docs.openclaw.ai
 
 ### Fixes
 
+- Agents/context pruning: guard assistant thinking/text char estimation against malformed blocks (missing `thinking`/`text` strings or null entries) so pruning no longer crashes with malformed provider content. (openclaw#35146) thanks @Sid-Qin.
 - Agents/schema cleaning: detect Venice + Grok model IDs as xAI-proxied targets so unsupported JSON Schema keywords are stripped before requests, preventing Venice/Grok `Invalid arguments` failures. (openclaw#35355) thanks @Sid-Qin.
 - Skills/native command deduplication: centralize skill command dedupe by canonical `skillName` in `listSkillCommandsForAgents` so duplicate suffixed variants (for example `_2`) are no longer surfaced across interfaces outside Discord. (#27521) thanks @shivama205.
 - Agents/xAI tool-call argument decoding: decode HTML-entity encoded xAI/Grok tool-call argument values (`&`, `"`, `<`, `>`, numeric entities) before tool execution so commands with shell operators and quotes no longer fail with parse errors. (#35276) Thanks @Sid-Qin.
diff --git a/src/agents/pi-extensions/context-pruning/pruner.test.ts b/src/agents/pi-extensions/context-pruning/pruner.test.ts
new file mode 100644
index 00000000000..3985bb2feb1
--- /dev/null
+++ b/src/agents/pi-extensions/context-pruning/pruner.test.ts
@@ -0,0 +1,112 @@
+import type { AgentMessage } from "@mariozechner/pi-agent-core";
+import type { ExtensionContext } from "@mariozechner/pi-coding-agent";
+import { describe, expect, it } from "vitest";
+import { pruneContextMessages } from "./pruner.js";
+import { DEFAULT_CONTEXT_PRUNING_SETTINGS } from "./settings.js";
+
+type AssistantMessage = Extract<AgentMessage, { role: "assistant" }>;
+type AssistantContentBlock = AssistantMessage["content"][number];
+
+const CONTEXT_WINDOW_1M = {
+  model: { contextWindow: 1_000_000 },
+} as unknown as ExtensionContext;
+
+function makeUser(text: string): AgentMessage {
+  return {
+    role: "user",
+    content: text,
+    timestamp: Date.now(),
+  };
+}
+
+function makeAssistant(content: AssistantMessage["content"]): AgentMessage {
+  return {
+    role: "assistant",
+    content,
+    api: "openai-responses",
+    provider: "openai",
+    model: "test-model",
+    usage: {
+      input: 1,
+      output: 1,
+      cacheRead: 0,
+      cacheWrite: 0,
+      totalTokens: 2,
+      cost: {
+        input: 0,
+        output: 0,
+        cacheRead: 0,
+        cacheWrite: 0,
+        total: 0,
+      },
+    },
+    stopReason: "stop",
+    timestamp: Date.now(),
+  };
+}
+
+describe("pruneContextMessages", () => {
+  it("does not crash on assistant message with malformed thinking block (missing thinking string)", () => {
+    const messages: AgentMessage[] = [
+      makeUser("hello"),
+      makeAssistant([
+        { type: "thinking" } as unknown as AssistantContentBlock,
+        { type: "text", text: "ok" },
+      ]),
+    ];
+    expect(() =>
+      pruneContextMessages({
+        messages,
+        settings: DEFAULT_CONTEXT_PRUNING_SETTINGS,
+        ctx: CONTEXT_WINDOW_1M,
+      }),
+    ).not.toThrow();
+  });
+
+  it("does not crash on assistant message with null content entries", () => {
+    const messages: AgentMessage[] = [
+      makeUser("hello"),
+      makeAssistant([null as unknown as AssistantContentBlock, { type: "text", text: "world" }]),
+    ];
+    expect(() =>
+      pruneContextMessages({
+        messages,
+        settings: DEFAULT_CONTEXT_PRUNING_SETTINGS,
+        ctx: CONTEXT_WINDOW_1M,
+      }),
+    ).not.toThrow();
+  });
+
+  it("does not crash on assistant message with malformed text block (missing text string)", () => {
+    const messages: AgentMessage[] = [
+      makeUser("hello"),
+      makeAssistant([
+        { type: "text" } as unknown as AssistantContentBlock,
+        { type: "thinking", thinking: "still fine" },
+      ]),
+    ];
+    expect(() =>
+      pruneContextMessages({
+        messages,
+        settings: DEFAULT_CONTEXT_PRUNING_SETTINGS,
+        ctx: CONTEXT_WINDOW_1M,
+      }),
+    ).not.toThrow();
+  });
+
+  it("handles well-formed thinking blocks correctly", () => {
+    const messages: AgentMessage[] = [
+      makeUser("hello"),
+      makeAssistant([
+        { type: "thinking", thinking: "let me think" },
+        { type: "text", text: "here is the answer" },
+      ]),
+    ];
+    const result = pruneContextMessages({
+      messages,
+      settings: DEFAULT_CONTEXT_PRUNING_SETTINGS,
+      ctx: CONTEXT_WINDOW_1M,
+    });
+    expect(result).toHaveLength(2);
+  });
+});
diff --git a/src/agents/pi-extensions/context-pruning/pruner.ts b/src/agents/pi-extensions/context-pruning/pruner.ts
index f9e3791b135..c195fa79e09 100644
--- a/src/agents/pi-extensions/context-pruning/pruner.ts
+++ b/src/agents/pi-extensions/context-pruning/pruner.ts
@@ -121,10 +121,13 @@ function estimateMessageChars(message: AgentMessage): number {
   if (message.role === "assistant") {
     let chars = 0;
     for (const b of message.content) {
-      if (b.type === "text") {
+      if (!b || typeof b !== "object") {
+        continue;
+      }
+      if (b.type === "text" && typeof b.text === "string") {
         chars += b.text.length;
       }
-      if (b.type === "thinking") {
+      if (b.type === "thinking" && typeof b.thinking === "string") {
         chars += b.thinking.length;
       }
       if (b.type === "toolCall") {