@codeworksh/aikit 0.1.0 → 0.2.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Sanchit Kudari
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -1,11 +1,11 @@
 # @codeworksh/aikit
 
-Bun-native AI toolkit for building AI agents and coding-agent workflows.
+Toolkit for Building AI Agents and Workflows.
 
 ## Installation
 
 ```bash
-bun add @codeworksh/aikit
+npm install @codeworksh/aikit
 ```
 
 ## Quick Start
@@ -70,7 +70,7 @@ const message = await stream.complete(
 
 ### Agent
 
-`agent.loop(...)` runs a full assistant turn loop on top of model streaming. It handles tool calls, validates arguments, mutates tool-call parts in-place, and emits higher-level agent events.
+`agent.run(...)` runs a full assistant turn loop on top of model streaming. It handles tool calls, validates arguments, mutates tool-call parts in-place, and emits higher-level agent events.
 
 ```ts
 import { Type } from "@sinclair/typebox";
@@ -80,7 +80,7 @@ const calculatorParams = Type.Object({
 	expression: Type.String(),
 });
 
-const calculatorTool: Agent.AgentTool<typeof calculatorParams> = {
+const calculatorTool = Agent.defineTool({
 	name: "calculator",
 	label: "Calculator",
 	description: "Evaluate a math expression",
@@ -94,7 +94,7 @@ const calculatorTool: Agent.AgentTool<typeof calculatorParams> = {
 			},
 		};
 	},
-};
+});
 
 const model = await llm("anthropic", "claude-haiku-4-5-20251001");
 if (!model) throw new Error("Missing model");
@@ -105,7 +105,7 @@ const prompt: Message.UserMessage = {
 	parts: [{ type: "text", text: "Use the calculator tool for 2 + 2." }],
 };
 
-const run = agent.loop(
+const run = agent.run(
 	{
 		model,
 		convertToLlm: async (messages) => messages,
@@ -137,7 +137,8 @@ console.log(messages.at(-1));
 
 Use `stream(...)` when you want raw provider events and manual control over tool handling.
 
-Use `agent.loop(...)` when you want a higher-level runtime that:
+Use `agent.run(...)` when you want a higher-level runtime that:
+
 - streams assistant message lifecycle events
 - validates tool arguments
 - executes tools sequentially or in parallel
@@ -192,7 +193,7 @@ const searchTool: Agent.AgentTool<typeof searchParams, { progress: number }, { h
 	description: "Search indexed documents",
 	parameters: searchParams,
 	async execute(callID, params, signal, onUpdate) {
-		onUpdate?.({
+		await onUpdate?.({
 			status: "running",
 			partial: {
 				content: [{ type: "text", text: `Searching for ${params.query}` }],
@@ -220,7 +221,7 @@ const prompt: Message.UserMessage = {
 	parts: [{ type: "text", text: "Search for agent loop documentation." }],
 };
 
-const run = agent.loop(
+const run = agent.run(
 	{
 		model,
 		apiKey: process.env.ANTHROPIC_API_KEY,
@@ -258,7 +259,7 @@ Tool execution mode is configurable with `toolExecution`:
 - `sequential`: prepares and executes each tool call one by one. Use this when tools depend on shared state, rate limits, or side effects that should not overlap.
 
 ```ts
-const run = agent.loop(
+const run = agent.run(
 	{
 		model,
 		convertToLlm: async (messages) => messages,
@@ -276,7 +277,7 @@ const run = agent.loop(
 Use it to inspect validated params, enforce policy, or block execution.
 
 ```ts
-const run = agent.loop(
+const run = agent.run(
 	{
 		model,
 		convertToLlm: async (messages) => messages,
@@ -303,7 +304,7 @@ Returning `{ block: true }` prevents the tool from running and turns that assist
 Use it to normalize output, attach metadata, or override the final tool result.
 
 ```ts
-const run = agent.loop(
+const run = agent.run(
 	{
 		model,
 		convertToLlm: async (messages) => messages,