diff --git a/src/agents/models-config.providers.ts b/src/agents/models-config.providers.ts
index f17f3b8ebdb..c6acf28977f 100644
--- a/src/agents/models-config.providers.ts
+++ b/src/agents/models-config.providers.ts
@@ -22,6 +22,7 @@ import {
   SYNTHETIC_BASE_URL,
   SYNTHETIC_MODEL_CATALOG,
 } from "./synthetic-models.js";
+import { OLLAMA_NATIVE_BASE_URL } from "./ollama-stream.js";
 import {
   TOGETHER_BASE_URL,
   TOGETHER_MODEL_CATALOG,
@@ -79,7 +80,7 @@ const QWEN_PORTAL_DEFAULT_COST = {
   cacheWrite: 0,
 };
 
-const OLLAMA_BASE_URL = "http://127.0.0.1:11434";
+const OLLAMA_BASE_URL = OLLAMA_NATIVE_BASE_URL;
 const OLLAMA_API_BASE_URL = OLLAMA_BASE_URL;
 const OLLAMA_DEFAULT_CONTEXT_WINDOW = 128000;
 const OLLAMA_DEFAULT_MAX_TOKENS = 8192;
diff --git a/src/agents/ollama-stream.test.ts b/src/agents/ollama-stream.test.ts
index 08f269c2c93..822fbc86fb4 100644
--- a/src/agents/ollama-stream.test.ts
+++ b/src/agents/ollama-stream.test.ts
@@ -49,12 +49,18 @@ describe("convertToOllamaMessages", () => {
     ]);
   });
 
-  it("converts tool result messages", () => {
+  it("converts tool result messages with 'tool' role", () => {
     const messages = [{ role: "tool", content: "file1.txt\nfile2.txt" }];
     const result = convertToOllamaMessages(messages);
     expect(result).toEqual([{ role: "tool", content: "file1.txt\nfile2.txt" }]);
   });
 
+  it("converts SDK 'toolResult' role to Ollama 'tool' role", () => {
+    const messages = [{ role: "toolResult", content: "command output here" }];
+    const result = convertToOllamaMessages(messages);
+    expect(result).toEqual([{ role: "tool", content: "command output here" }]);
+  });
+
   it("handles empty messages array", () => {
     const result = convertToOllamaMessages([]);
     expect(result).toEqual([]);
@@ -99,11 +105,12 @@ describe("buildAssistantMessage", () => {
     };
     const result = buildAssistantMessage(response, modelInfo);
     expect(result.stopReason).toBe("end_turn");
-    expect(result.content.length).toBe(2); // empty text + tool_use
-    expect(result.content[1].type).toBe("tool_use");
-    const toolUse = result.content[1] as { type: "tool_use"; name: string; input: Record<string, unknown> };
+    expect(result.content.length).toBe(1); // tool_use only (empty content is skipped)
+    expect(result.content[0].type).toBe("tool_use");
+    const toolUse = result.content[0] as { type: "tool_use"; id: string; name: string; input: Record<string, unknown> };
     expect(toolUse.name).toBe("bash");
     expect(toolUse.input).toEqual({ command: "ls -la" });
+    expect(toolUse.id).toMatch(/^ollama_call_[0-9a-f-]{36}$/);
   });
 
   it("sets all costs to zero for local models", () => {
diff --git a/src/agents/ollama-stream.ts b/src/agents/ollama-stream.ts
index 8697de6b629..6c7c9acdb23 100644
--- a/src/agents/ollama-stream.ts
+++ b/src/agents/ollama-stream.ts
@@ -1,5 +1,4 @@
 import type { StreamFn } from "@mariozechner/pi-agent-core";
-import type { SimpleStreamOptions } from "@mariozechner/pi-ai";
 import { AssistantMessageEventStream } from "@mariozechner/pi-ai";
 
 // ── Ollama /api/chat request types ──────────────────────────────────────────
@@ -137,7 +136,9 @@ export function convertToOllamaMessages(
         content: text,
         ...(toolCalls.length > 0 ? { tool_calls: toolCalls } : {}),
       });
-    } else if (role === "tool") {
+    } else if (role === "tool" || role === "toolResult") {
+      // SDK uses "toolResult" (camelCase) for tool result messages.
+      // Ollama API expects "tool" role.
       const text = extractTextContent(msg.content);
       result.push({ role: "tool", content: text });
     }
@@ -146,6 +147,39 @@ export function convertToOllamaMessages(
   return result;
 }
 
+// ── Tool extraction ─────────────────────────────────────────────────────────
+
+type SdkToolDef = {
+  name?: string;
+  description?: string;
+  parameters?: Record<string, unknown>;
+  inputSchema?: Record<string, unknown>;
+  [key: string]: unknown;
+};
+
+function extractOllamaTools(tools: unknown[] | undefined): OllamaTool[] {
+  if (!tools || !Array.isArray(tools)) {
+    return [];
+  }
+  const result: OllamaTool[] = [];
+  for (const tool of tools) {
+    const t = tool as SdkToolDef;
+    const name = t.name;
+    if (typeof name !== "string" || !name) {
+      continue;
+    }
+    result.push({
+      type: "function",
+      function: {
+        name,
+        description: typeof t.description === "string" ? t.description : "",
+        parameters: t.inputSchema ?? t.parameters ?? {},
+      },
+    });
+  }
+  return result;
+}
+
 // ── Response conversion ─────────────────────────────────────────────────────
 
 interface AssistantMessageLike {
@@ -172,8 +206,6 @@ interface AssistantMessageLike {
   timestamp: number;
 }
 
-let toolCallIdCounter = 0;
-
 export function buildAssistantMessage(
   response: OllamaChatResponse,
   modelInfo: { api: string; provider: string; id: string },
@@ -187,10 +219,9 @@ export function buildAssistantMessage(
   const toolCalls = response.message.tool_calls;
   if (toolCalls && toolCalls.length > 0) {
     for (const tc of toolCalls) {
-      toolCallIdCounter += 1;
       content.push({
         type: "tool_use",
-        id: `ollama_call_${toolCallIdCounter}_${Date.now()}`,
+        id: `ollama_call_${crypto.randomUUID()}`,
         name: tc.function.name,
         input: tc.function.arguments,
       });
@@ -244,7 +275,7 @@ export async function* parseNdjsonStream(
       try {
         yield JSON.parse(trimmed) as OllamaChatResponse;
       } catch {
-        // Skip malformed lines
+        console.warn("[ollama-stream] Skipping malformed NDJSON line:", trimmed.slice(0, 120));
       }
     }
   }
@@ -253,7 +284,7 @@ export async function* parseNdjsonStream(
     try {
       yield JSON.parse(buffer.trim()) as OllamaChatResponse;
     } catch {
-      // Skip malformed trailing data
+      console.warn("[ollama-stream] Skipping malformed trailing data:", buffer.trim().slice(0, 120));
     }
   }
 }
@@ -268,13 +299,17 @@ export function createOllamaStreamFn(baseUrl: string): StreamFn {
 
     const run = async () => {
       try {
-        const ctx = context as { messages?: AgentMessage[]; system?: string };
+        const ctx = context as { messages?: AgentMessage[]; system?: string; tools?: unknown[] };
         const ollamaMessages = convertToOllamaMessages(ctx.messages ?? [], ctx.system as string);
 
+        // Extract tools from context if available and convert to Ollama format.
+        const ollamaTools = extractOllamaTools(ctx.tools);
+
         const body: OllamaChatRequest = {
           model: model.id,
           messages: ollamaMessages,
           stream: true,
+          ...(ollamaTools.length > 0 ? { tools: ollamaTools } : {}),
           ...(typeof options?.temperature === "number"
             ? { options: { temperature: options.temperature } }
             : {}),