diff --git a/src/index.ts b/src/index.ts
index c769e99..228cd8e 100644
--- a/src/index.ts
+++ b/src/index.ts
@@ -9,6 +9,7 @@ import { contextRegistry } from "./introspection/contextRegistry";
 import { ModelRouter } from "./router/modelRouter";
 import { ModelRouterOptions, ApiKeyConfig } from "./router/types";
 import { apiKeyManager } from "./router/apiKeyManager";
+import { taskRegistry, TaskRegistry } from "./router/taskRegistry";
 
 let globalBudgetManager: BudgetManager | null = null;
 let globalModelRouter: ModelRouter | null = null;
@@ -227,6 +228,7 @@ export async function listModels(options: Omit<ListModelsOptions, 'budgetManager
 
 // Keep the original export for backward compatibility
 export { listAvailableModels };
+export { taskRegistry, TaskRegistry };
 
 // Export types for TypeScript users
 export type {
@@ -252,6 +254,11 @@ export type {
   ApiKeyConfig
 } from "./router/types";
 
+export type {
+  TaskConfiguration,
+  TaskRegistration
+} from "./router/taskRegistry";
+
 /**
  * Model configuration for bulk registration
  */
diff --git a/src/router/taskRegistry.ts b/src/router/taskRegistry.ts
new file mode 100644
index 0000000..64298ef
--- /dev/null
+++ b/src/router/taskRegistry.ts
@@ -0,0 +1,229 @@
+/**
+ * Task configuration registry for Smart Model Selection.
+ */
+
+export interface TaskConfiguration {
+  /** Stable task identifier, for example "code_generation". */
+  taskType: string;
+  /** Preferred model for this task. */
+  model: string;
+  /** Human-readable reason for selecting the model. */
+  reason: string;
+  /** Keyword hints used by lightweight classifiers. */
+  keywords?: string[];
+  /** Regex patterns used by pattern detectors. */
+  patterns?: RegExp[];
+  /** Higher-priority tasks are evaluated first. */
+  priority?: number;
+  /** Optional token threshold for context-sensitive tasks. */
+  contextThreshold?: number;
+}
+
+export type TaskRegistration = Omit<TaskConfiguration, "taskType"> & {
+  taskType?: string;
+};
+
+const BUILT_IN_TASKS: TaskConfiguration[] = [
+  {
+    taskType: "code_generation",
+    model: "claude-3-5-sonnet-20241022",
+    reason: "Claude excels at producing maintainable code",
+    keywords: ["write code", "create function", "implement", "build", "develop", "program"],
+    patterns: [/write.*code/i, /create.*function/i, /implement.*class/i],
+    priority: 100,
+  },
+  {
+    taskType: "code_review",
+    model: "claude-3-5-sonnet-20241022",
+    reason: "Claude is strong at code review, debugging, and refactoring",
+    keywords: ["review code", "find bugs", "optimize", "refactor", "improve code", "debug"],
+    patterns: [/review.*code/i, /find.*bug/i, /refactor/i, /optimize/i],
+    priority: 95,
+  },
+  {
+    taskType: "math_reasoning",
+    model: "o1-mini",
+    reason: "o1-mini is optimized for mathematical reasoning",
+    keywords: ["calculate", "solve", "compute", "equation", "formula", "math"],
+    patterns: [/solve.*equation/i, /calculate/i, /mathematical/i],
+    priority: 90,
+  },
+  {
+    taskType: "complex_reasoning",
+    model: "o1",
+    reason: "o1 is suited to deeper logic and step-by-step reasoning",
+    keywords: ["analyze", "reason", "logic", "deduce", "infer", "prove", "derive"],
+    patterns: [/step.*by.*step/i, /reasoning/i, /logical.*analysis/i],
+    priority: 85,
+  },
+  {
+    taskType: "document_analysis",
+    model: "gemini-2.5-pro",
+    reason: "Gemini has a very large context window for long documents",
+    keywords: ["summarize document", "analyze document", "extract from", "review document"],
+    patterns: [/summarize.*document/i, /analyze.*pdf/i, /extract.*information/i],
+    priority: 80,
+    contextThreshold: 50000,
+  },
+  {
+    taskType: "creative_writing",
+    model: "gpt-4o",
+    reason: "GPT-4o is effective for creative and engaging prose",
+    keywords: ["write story", "create content", "blog post", "article", "creative"],
+    patterns: [/write.*story/i, /creative.*writing/i, /blog.*post/i],
+    priority: 75,
+  },
+  {
+    taskType: "technical_documentation",
+    model: "claude-3-5-sonnet-20241022",
+    reason: "Claude is strong at technical writing and documentation",
+    keywords: ["document", "documentation", "api docs", "technical writing"],
+    patterns: [/write.*documentation/i, /create.*docs/i],
+    priority: 72,
+  },
+  {
+    taskType: "translation",
+    model: "gpt-4o-mini",
+    reason: "Translation is usually handled well by a cost-effective model",
+    keywords: ["translate", "translation", "convert to"],
+    patterns: [/translate.*to/i, /translation/i],
+    priority: 70,
+  },
+  {
+    taskType: "simple_chat",
+    model: "gpt-4o-mini",
+    reason: "GPT-4o-mini is fast and inexpensive for simple conversation",
+    keywords: ["hello", "hi", "how are you", "thanks", "thank you", "help"],
+    patterns: [/^(hi|hello|hey)/i, /how.*are.*you/i, /thank/i],
+    priority: 65,
+  },
+  {
+    taskType: "data_extraction",
+    model: "gpt-4o-mini",
+    reason: "Structured extraction works well on a cost-effective model",
+    keywords: ["extract", "parse", "get data from", "scrape", "pull data"],
+    patterns: [/extract.*from/i, /parse.*json/i, /get.*data/i],
+    priority: 60,
+  },
+  {
+    taskType: "factual_qa",
+    model: "gpt-4o-mini",
+    reason: "GPT-4o-mini is efficient for straightforward factual answers",
+    keywords: ["what is", "who is", "when did", "where is", "how many"],
+    patterns: [/^(what|who|when|where|how|why)\b/i],
+    priority: 55,
+  },
+  {
+    taskType: "chinese_language",
+    model: "moonshot-v1-32k",
+    reason: "Kimi is optimized for Chinese language understanding",
+    keywords: ["chinese"],
+    patterns: [/[\u4e00-\u9fff]/],
+    priority: 98,
+  },
+];
+
+/**
+ * Stores built-in and custom smart-routing task configurations.
+ */
+export class TaskRegistry {
+  private tasks: Map<string, TaskConfiguration> = new Map();
+
+  constructor(initialTasks: TaskConfiguration[] = BUILT_IN_TASKS) {
+    for (const task of initialTasks) {
+      this.registerTask(task.taskType, task);
+    }
+  }
+
+  /**
+   * Register or replace a task configuration.
+   */
+  public registerTask(taskType: string, config: TaskRegistration): void {
+    const normalizedTaskType = this.normalizeTaskType(taskType);
+    const taskConfig: TaskConfiguration = {
+      ...config,
+      taskType: normalizedTaskType,
+    };
+
+    this.validateTask(taskConfig);
+    this.tasks.set(normalizedTaskType, this.cloneTask(taskConfig));
+  }
+
+  /**
+   * Get a task configuration by type.
+   */
+  public getTask(taskType: string): TaskConfiguration | undefined {
+    const task = this.tasks.get(this.normalizeTaskType(taskType));
+    return task ? this.cloneTask(task) : undefined;
+  }
+
+  /**
+   * Get all tasks sorted by priority, highest first.
+   */
+  public getAllTasks(): TaskConfiguration[] {
+    return Array.from(this.tasks.values())
+      .map(task => this.cloneTask(task))
+      .sort((a, b) => (b.priority ?? 0) - (a.priority ?? 0));
+  }
+
+  /**
+   * Check whether a task type is registered.
+   */
+  public hasTask(taskType: string): boolean {
+    return this.tasks.has(this.normalizeTaskType(taskType));
+  }
+
+  /**
+   * Validate a task configuration.
+   */
+  public validateTask(task: TaskConfiguration): void {
+    if (!task.taskType || typeof task.taskType !== "string" || task.taskType.trim() === "") {
+      throw new Error("TokenFirewall TaskRegistry: taskType must be a non-empty string");
+    }
+
+    if (!task.model || typeof task.model !== "string" || task.model.trim() === "") {
+      throw new Error(`TokenFirewall TaskRegistry: model must be a non-empty string for task "${task.taskType}"`);
+    }
+
+    if (!task.reason || typeof task.reason !== "string" || task.reason.trim() === "") {
+      throw new Error(`TokenFirewall TaskRegistry: reason must be a non-empty string for task "${task.taskType}"`);
+    }
+
+    if (task.keywords !== undefined && !this.isStringArray(task.keywords)) {
+      throw new Error(`TokenFirewall TaskRegistry: keywords must be an array of strings for task "${task.taskType}"`);
+    }
+
+    if (task.patterns !== undefined && !task.patterns.every(pattern => pattern instanceof RegExp)) {
+      throw new Error(`TokenFirewall TaskRegistry: patterns must be an array of RegExp instances for task "${task.taskType}"`);
+    }
+
+    if (task.priority !== undefined && (!Number.isFinite(task.priority) || task.priority < 0)) {
+      throw new Error(`TokenFirewall TaskRegistry: priority must be a non-negative number for task "${task.taskType}"`);
+    }
+
+    if (task.contextThreshold !== undefined && (!Number.isFinite(task.contextThreshold) || task.contextThreshold < 0)) {
+      throw new Error(
+        `TokenFirewall TaskRegistry: contextThreshold must be a non-negative number for task "${task.taskType}"`
+      );
+    }
+  }
+
+  private normalizeTaskType(taskType: string): string {
+    return taskType.trim().toLowerCase();
+  }
+
+  private isStringArray(value: unknown): value is string[] {
+    return Array.isArray(value) && value.every(item => typeof item === "string" && item.trim() !== "");
+  }
+
+  private cloneTask(task: TaskConfiguration): TaskConfiguration {
+    return {
+      ...task,
+      keywords: task.keywords ? [...task.keywords] : undefined,
+      patterns: task.patterns ? [...task.patterns] : undefined,
+    };
+  }
+}
+
+export const taskRegistry = new TaskRegistry();
+export const builtInTasks = BUILT_IN_TASKS.map(task => ({ ...task }));
diff --git a/tests/task-registry.test.js b/tests/task-registry.test.js
new file mode 100644
index 0000000..8f550ae
--- /dev/null
+++ b/tests/task-registry.test.js
@@ -0,0 +1,112 @@
+/**
+ * Task registry tests.
+ *
+ * Run: npm run build && node tests/task-registry.test.js
+ */
+
+const assert = require("assert");
+const { TaskRegistry, taskRegistry } = require("../dist/index.js");
+
+function test(name, fn) {
+  try {
+    fn();
+    console.log(`✓ ${name}`);
+  } catch (error) {
+    console.error(`✗ ${name}`);
+    throw error;
+  }
+}
+
+test("registers all built-in smart routing tasks", () => {
+  const tasks = taskRegistry.getAllTasks();
+  const taskTypes = tasks.map(task => task.taskType);
+
+  assert.ok(tasks.length >= 12);
+  assert.ok(taskTypes.includes("code_generation"));
+  assert.ok(taskTypes.includes("document_analysis"));
+  assert.ok(taskTypes.includes("chinese_language"));
+  assert.ok(taskTypes.includes("technical_documentation"));
+});
+
+test("returns tasks ordered by priority", () => {
+  const registry = new TaskRegistry([
+    {
+      taskType: "low_priority",
+      model: "gpt-4o-mini",
+      reason: "Low priority",
+      priority: 1,
+    },
+    {
+      taskType: "high_priority",
+      model: "gpt-4o",
+      reason: "High priority",
+      priority: 99,
+    },
+  ]);
+
+  assert.deepStrictEqual(
+    registry.getAllTasks().map(task => task.taskType),
+    ["high_priority", "low_priority"]
+  );
+});
+
+test("supports custom task registration", () => {
+  const registry = new TaskRegistry([]);
+
+  registry.registerTask("legal_analysis", {
+    model: "gpt-4o",
+    reason: "Complex legal reasoning",
+    keywords: ["legal", "contract", "clause"],
+    patterns: [/legal.*analysis/i],
+    priority: 8,
+  });
+
+  const task = registry.getTask("legal_analysis");
+  assert.ok(registry.hasTask("legal_analysis"));
+  assert.strictEqual(task.model, "gpt-4o");
+  assert.strictEqual(task.patterns[0].test("legal contract analysis"), true);
+});
+
+test("defensively copies returned task arrays", () => {
+  const registry = new TaskRegistry([]);
+
+  registry.registerTask("simple", {
+    model: "gpt-4o-mini",
+    reason: "Simple test task",
+    keywords: ["hello"],
+    patterns: [/hello/i],
+  });
+
+  const task = registry.getTask("simple");
+  task.keywords.push("mutated");
+  task.patterns.push(/mutated/i);
+
+  const freshTask = registry.getTask("simple");
+  assert.deepStrictEqual(freshTask.keywords, ["hello"]);
+  assert.strictEqual(freshTask.patterns.length, 1);
+});
+
+test("validates required task fields", () => {
+  const registry = new TaskRegistry([]);
+
+  assert.throws(
+    () => registry.registerTask("bad", { model: "", reason: "missing model" }),
+    /model must be a non-empty string/
+  );
+
+  assert.throws(
+    () => registry.registerTask("bad", { model: "gpt-4o", reason: "" }),
+    /reason must be a non-empty string/
+  );
+
+  assert.throws(
+    () => registry.registerTask("bad", {
+      model: "gpt-4o",
+      reason: "invalid patterns",
+      patterns: ["not-regex"],
+    }),
+    /patterns must be an array of RegExp/
+  );
+});
+
+console.log("Task registry tests passed.");