feat(chat): decouple primitives from LangGraph via runtime-neutral ChatAgent contract

.github/workflows/ci.yml

@@ -17,9 +17,9 @@ jobs:
           node-version: 22
           cache: npm
       - run: npm ci
-      - run: npx nx lint agent
-      - run: npx nx test agent --coverage
-      - run: npx nx build agent --configuration=production
+      - run: npx nx lint langgraph
+      - run: npx nx test langgraph --coverage
+      - run: npx nx build langgraph --configuration=production
 
   website:
     name: Website — lint / build

.github/workflows/publish.yml

@@ -17,8 +17,8 @@ jobs:
           registry-url: https://registry.npmjs.org
       - run: npm ci
       - run: npx nx test mcp --skip-nx-cache
-      - run: npx nx test agent
-      - run: npx nx build agent --configuration=production
+      - run: npx nx test langgraph
+      - run: npx nx build langgraph --configuration=production
       - name: Publish to npm
         run: npx nx-release-publish agent
         env:

AGENTS.md

@@ -36,9 +36,9 @@ This file is for agents working in this repository. It is contributor-facing, no
 
 ## Repo Layout
 
-- `libs/agent`: main Angular library (`@cacheplane/angular`).
+- `libs/langgraph`: main Angular library (`@cacheplane/langgraph`).
 - `apps/website`: docs and marketing site.
-- `packages/mcp`: MCP server package (`@cacheplane/angular-mcp`).
+- `packages/mcp`: MCP server package (`@cacheplane/langgraph-mcp`).
 - `e2e/agent-e2e`: end-to-end coverage for the workspace.
 - `apps/demo` and `apps/demo-e2e`: demo application and related end-to-end coverage.
 

COMMERCIAL.md

@@ -1,6 +1,6 @@
 # Commercial Licensing
 
-`@cacheplane/angular` is source-available software dual-licensed under:
+`@cacheplane/langgraph` is source-available software dual-licensed under:
 
 - **PolyForm Noncommercial 1.0.0** — free for noncommercial use (see [`LICENSE`](./LICENSE))
 - **Angular Agent Framework Commercial License** — required for commercial use (see [`LICENSE-COMMERCIAL`](./LICENSE-COMMERCIAL))

LICENSE-COMMERCIAL

@@ -3,7 +3,7 @@ Commercial License
 Copyright (c) 2026 Brian Love d/b/a cacheplane. All rights reserved.
 
 This Commercial License ("License") governs commercial use of the
-@cacheplane/angular software ("Software"). Use of the Software for
+@cacheplane/langgraph software ("Software"). Use of the Software for
 commercial purposes requires a valid license purchased from cacheplane.
 
 --- LICENSE TIERS ---

README.md

@@ -11,7 +11,7 @@
 </p>
 
 <p align="center">
-  <a href="https://www.npmjs.com/package/@cacheplane/angular">
+  <a href="https://www.npmjs.com/package/@cacheplane/langgraph">
     <img alt="npm version" src="https://img.shields.io/npm/v/@cacheplane%2Fangular?color=6C8EFF&labelColor=080B14&style=flat-square" />
   </a>
   <a href="./LICENSE">
@@ -34,7 +34,7 @@
 ## Install
 
 ```bash
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 ```
 
 **Peer dependencies:** `@angular/core ^20.0.0 || ^21.0.0`, `@langchain/core ^1.1.0`, `@langchain/langgraph-sdk ^1.7.0`, `rxjs ~7.8.0`
@@ -45,7 +45,7 @@ npm install @cacheplane/angular
 
 ```typescript
 import { Component } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({
@@ -145,7 +145,7 @@ That's it. `chat.messages()` is an Angular Signal. Bind it directly in your temp
 
 ## License
 
-`@cacheplane/angular` is source-available software dual-licensed:
+`@cacheplane/langgraph` is source-available software dual-licensed:
 
 - **PolyForm Noncommercial 1.0.0** — free for noncommercial use (personal projects, academic, research, non-profit internal tooling). See [`LICENSE`](./LICENSE).
 - **Angular Agent Framework Commercial License** — required for any for-profit or revenue-generating use. See [`LICENSE-COMMERCIAL`](./LICENSE-COMMERCIAL) and [`COMMERCIAL.md`](./COMMERCIAL.md).

apps/demo/src/app/chat-demo/chat-demo.component.ts

@@ -1,5 +1,5 @@
 import { Component, Input, OnInit, Injector, runInInjectionContext } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({

apps/website/content/AGENTS.md.template

@@ -3,21 +3,21 @@
 Angular agent framework for LangChain/LangGraph. Provides `agent()` — Signal-native streaming for Angular agents, built for LangGraph.
 
 ## Install
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 
 ## Key requirement
 `agent()` MUST be called within an Angular injection context (component constructor or field initializer). Calling it in ngOnInit or any async context throws "NG0203: inject() must be called from an injection context".
 
 ## Basic usage
 ```typescript
 // app.config.ts
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 export const appConfig: ApplicationConfig = {
   providers: [provideAgent({ apiUrl: 'http://localhost:2024' })]
 };
 
 // chat.component.ts
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({ template: `
@@ -38,7 +38,7 @@ export class ChatComponent {
 
 ## MCP server (for tool access)
 Add to ~/.claude/settings.json:
-{"mcpServers":{"angular-agent":{"command":"npx","args":["@cacheplane/angular-mcp"]}}}
+{"mcpServers":{"angular-agent":{"command":"npx","args":["@cacheplane/langgraph-mcp"]}}}
 
 ## Version check
 If this file is stale, fetch the latest: https://cacheplane.ai/llms-full.txt

apps/website/content/CLAUDE.md.template

@@ -3,21 +3,21 @@
 Angular agent framework for LangChain/LangGraph. Provides `agent()` — Signal-native streaming for Angular agents, built for LangGraph.
 
 ## Install
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 
 ## Key requirement
 `agent()` MUST be called within an Angular injection context (component constructor or field initializer). Calling it in ngOnInit or any async context throws "NG0203: inject() must be called from an injection context".
 
 ## Basic usage
 ```typescript
 // app.config.ts
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 export const appConfig: ApplicationConfig = {
   providers: [provideAgent({ apiUrl: 'http://localhost:2024' })]
 };
 
 // chat.component.ts
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({ template: `
@@ -38,7 +38,7 @@ export class ChatComponent {
 
 ## MCP server (for tool access)
 Add to ~/.claude/settings.json:
-{"mcpServers":{"angular-agent":{"command":"npx","args":["@cacheplane/angular-mcp"]}}}
+{"mcpServers":{"angular-agent":{"command":"npx","args":["@cacheplane/langgraph-mcp"]}}}
 
 ## Version check
 If this file is stale, fetch the latest: https://cacheplane.ai/llms-full.txt

apps/website/content/docs/agent/api/agent.mdx

@@ -5,7 +5,7 @@
 When the `url` signal changes, the resource tears down the previous connection and opens a fresh one automatically. You never write subscription management or cleanup logic yourself.
 
 ```ts
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 // Inside a component or service with injection context
 const repo = agent<Repository>({

apps/website/content/docs/agent/api/fetch-stream-transport.mdx

@@ -11,7 +11,7 @@ In most apps you will never import or inject `FetchStreamTransport` by name —
 
 ```ts
 import { inject } from '@angular/core';
-import { agent, FetchStreamTransport } from '@cacheplane/angular';
+import { agent, FetchStreamTransport } from '@cacheplane/langgraph';
 
 // Override transport for a single resource
 const events = agent<Event>({

apps/website/content/docs/agent/api/mock-stream-transport.mdx

@@ -13,7 +13,7 @@ import {
   provideAgent,
   MockAgentTransport,
   agent,
-} from '@cacheplane/angular';
+} from '@cacheplane/langgraph';
 
 @Component({ template: '' })
 class RepoComponent {

apps/website/content/docs/agent/api/provide-agent.mdx

@@ -9,7 +9,7 @@ import { bootstrapApplication } from '@angular/platform-browser';
 import {
   provideAgent,
   FetchStreamTransport,
-} from '@cacheplane/angular';
+} from '@cacheplane/langgraph';
 import { AppComponent } from './app/app.component';
 
 bootstrapApplication(AppComponent, {

apps/website/content/docs/agent/concepts/agent-architecture.mdx

@@ -1,6 +1,6 @@
 # Agent Architecture
 
-How AI agents work — the planning, execution, and tool-calling lifecycle that agent() connects your Angular app to. This page shows you the Python patterns that power modern agents and exactly how each pattern surfaces in Angular through `@cacheplane/angular`.
+How AI agents work — the planning, execution, and tool-calling lifecycle that agent() connects your Angular app to. This page shows you the Python patterns that power modern agents and exactly how each pattern surfaces in Angular through `@cacheplane/langgraph`.
 
 <Callout type="info" title="Python + Angular, both sides">
 Every section below shows the Python backend code first, then the Angular frontend code that consumes it. You need both halves to build a production agent application — LangGraph handles the intelligence, agent() handles the reactivity.
@@ -149,7 +149,7 @@ graph = builder.compile()
 
 ```typescript
 import { ChangeDetectionStrategy, Component, computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 interface AgentState {
   messages: BaseMessage[];
@@ -401,7 +401,7 @@ graph = builder.compile()
 
 ```typescript
 import { ChangeDetectionStrategy, Component, computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 interface OrchestratorState {
   messages: BaseMessage[];

apps/website/content/docs/agent/concepts/angular-signals.mdx

@@ -60,7 +60,7 @@ const status = toSignal(status$, { initialValue: 'idle' });
 <Tab label="What you see">
 
 ```typescript
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 // You never touch BehaviorSubjects or toSignal() yourself.
 // agent() hands you clean Signals:
@@ -154,7 +154,7 @@ console.log(chat.isLoading());  // false
 
 ```typescript
 import { computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 const chat = agent<ChatState>({
   assistantId: 'chat_agent',
@@ -259,7 +259,7 @@ Angular's new control flow syntax (`@if`, `@for`, `@switch`) works naturally wit
 
 ```typescript
 import { ChangeDetectionStrategy, Component, computed, effect, ElementRef, ViewChild } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 @Component({
   selector: 'app-chat',
@@ -457,7 +457,7 @@ graph = builder.compile()
 
 ```typescript
 import { ChangeDetectionStrategy, Component, computed, effect } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 @Component({
   selector: 'app-chat',

apps/website/content/docs/agent/concepts/state-management.mdx

@@ -17,7 +17,7 @@ This inversion is intentional. Agent state can span multiple LLM calls, tool exe
 Your Angular component calls `agent.submit({ messages: [userMsg] })`. No state is stored in the component.
 </Step>
 <Step title="agent() sends the request">
-`@cacheplane/angular` forwards the input to `FetchStreamTransport`, which opens an HTTP POST and SSE connection to LangGraph Platform.
+`@cacheplane/langgraph` forwards the input to `FetchStreamTransport`, which opens an HTTP POST and SSE connection to LangGraph Platform.
 </Step>
 <Step title="LangGraph Platform executes the graph">
 The agent runs its nodes — calling the LLM, invoking tools, checking conditions — and streams SSE events back with incremental state updates.
@@ -419,7 +419,7 @@ def researcher_node(state: ResearchState) -> dict:
 
 ```typescript
 import { BaseMessage } from '@langchain/core/messages';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 interface ResearchState {
   messages: BaseMessage[];

apps/website/content/docs/agent/getting-started/installation.mdx

@@ -19,7 +19,7 @@ A running LangGraph agent accessible via HTTP. Can be local (langgraph dev) or d
 ## Install the package
 
 ```bash
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 ```
 
 This installs the library and its peer dependencies including `@langchain/langgraph-sdk`.
@@ -31,7 +31,7 @@ Add `provideAgent()` to your application configuration. This sets global default
 ```typescript
 // app.config.ts
 import { ApplicationConfig } from '@angular/core';
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 import { environment } from '../environments/environment';
 
 export const appConfig: ApplicationConfig = {

apps/website/content/docs/agent/getting-started/introduction.mdx

@@ -126,7 +126,7 @@ Now connect your Angular app to the running agent using agent().
 <Step title="Install the package">
 
 ```bash
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 ```
 
 </Step>
@@ -135,7 +135,7 @@ npm install @cacheplane/angular
 ```typescript
 // app.config.ts
 import { ApplicationConfig } from '@angular/core';
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 
 export const appConfig: ApplicationConfig = {
   providers: [
@@ -155,7 +155,7 @@ export const appConfig: ApplicationConfig = {
 ```typescript
 // chat.component.ts
 import { Component, signal, computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({

apps/website/content/docs/agent/getting-started/quickstart.mdx

@@ -10,7 +10,7 @@ Angular 20+ project with Node.js 18+. If you need setup help, see the [Installat
 <Step title="Install the package">
 
 ```bash
-npm install @cacheplane/angular
+npm install @cacheplane/langgraph
 ```
 
 </Step>
@@ -20,7 +20,7 @@ Add `provideAgent()` to your application config with your LangGraph Platform URL
 
 ```typescript
 // app.config.ts
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 
 export const appConfig: ApplicationConfig = {
   providers: [
@@ -42,7 +42,7 @@ Use `agent()` in a component field initializer. Every property on the returned r
 ```typescript
 // chat.component.ts
 import { Component, ChangeDetectionStrategy, signal, computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 import type { BaseMessage } from '@langchain/core/messages';
 
 @Component({

apps/website/content/docs/agent/guides/deployment.mdx

@@ -108,7 +108,7 @@ export const environment = {
 Wire the environment into `provideAgent()`:
 
 ```typescript
-import { provideAgent } from '@cacheplane/angular';
+import { provideAgent } from '@cacheplane/langgraph';
 import { environment } from '../environments/environment';
 
 export const appConfig: ApplicationConfig = {
@@ -202,7 +202,7 @@ Production apps need graceful error handling. Build a reactive error boundary us
 
 ```typescript
 import { ChangeDetectionStrategy, Component, computed } from '@angular/core';
-import { agent } from '@cacheplane/angular';
+import { agent } from '@cacheplane/langgraph';
 
 @Component({
   selector: 'app-chat',

apps/website/content/docs/agent/guides/interrupts.mdx

@@ -143,7 +143,7 @@ import {
   signal,
   ChangeDetectionStrategy,
 } from '@angular/core';
-import { agent, BaseMessage } from '@cacheplane/angular';
+import { agent, BaseMessage } from '@cacheplane/langgraph';
 
 interface ApprovalPayload {
   action: string;
@@ -354,7 +354,7 @@ import {
   computed,
   ChangeDetectionStrategy,
 } from '@angular/core';
-import { agent, BaseMessage } from '@cacheplane/angular';
+import { agent, BaseMessage } from '@cacheplane/langgraph';
 
 interface StepApproval {
   step_number: number;
@@ -454,7 +454,7 @@ By default, `interrupt()` returns an untyped object. The BagTemplate generic par
 BagTemplate is a type parameter on the agent configuration that maps signal names to their types. When you specify an interrupt type through BagTemplate, the `interrupt()` signal returns a properly typed object instead of `unknown`. This means your template expressions, computed signals, and event handlers all benefit from compile-time checking.
 
 ```typescript
-import { agent, BagTemplate } from '@cacheplane/angular';
+import { agent, BagTemplate } from '@cacheplane/langgraph';
 
 // Define the exact shape of your interrupt payload
 interface DeployApproval {