@auto-engineer/model-factory 1.148.0 → 1.149.0

package/.turbo/turbo-build.log

@@ -1,5 +1,5 @@
 
-> @auto-engineer/model-factory@1.148.0 build /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
+> @auto-engineer/model-factory@1.149.0 build /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
 > tsc && tsx ../../scripts/fix-esm-imports.ts
 
 Fixed ESM imports in dist/

package/.turbo/turbo-test.log

@@ -1,5 +1,5 @@
 
-> @auto-engineer/model-factory@1.147.0 test /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
+> @auto-engineer/model-factory@1.148.0 test /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
 > vitest run --reporter=dot
 
 
@@ -9,6 +9,6 @@
 
  Test Files  1 passed (1)
       Tests  5 passed (5)
-   Start at  21:35:18
-   Duration  2.14s (transform 361ms, setup 0ms, collect 272ms, tests 162ms, environment 0ms, prepare 645ms)
+   Start at  07:22:24
+   Duration  2.21s (transform 375ms, setup 0ms, collect 300ms, tests 154ms, environment 0ms, prepare 661ms)
 

package/.turbo/turbo-type-check.log

@@ -1,4 +1,4 @@
 
-> @auto-engineer/model-factory@1.147.0 type-check /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
+> @auto-engineer/model-factory@1.148.0 type-check /home/runner/work/auto-engineer/auto-engineer/packages/model-factory
 > tsc --noEmit
 

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # @auto-engineer/model-factory
 
+## 1.149.0
+
+### Minor Changes
+
+- [`e1eebbd`](https://github.com/BeOnAuto/auto-engineer/commit/e1eebbdf4f209780e790094d2e6887c4fa809f98) Thanks [@github-actions[bot]](https://github.com/github-actions%5Bbot%5D)! - - **server-generator-apollo-emmett**: add Given state ref hints to state.ts.ejs
+  - **server-generator-apollo-emmett**: context-aware nonCommandField instructions
+  - **server-generator-apollo-emmett**: add state context instruction
+  - **server-generator-apollo-emmett**: extract shared template helpers
+  - **server-generator-apollo-emmett**: filter state refs from hasGivenEvents in decide.ts.ejs
+
+### Patch Changes
+
+- [`d38c81e`](https://github.com/BeOnAuto/auto-engineer/commit/d38c81e7bb442a39626564cf4f6d8d55b60d0a38) Thanks [@SamHatoum](https://github.com/SamHatoum)! -
+
 ## 1.148.0
 
 ### Minor Changes

package/README.md

@@ -0,0 +1,171 @@
+# @auto-engineer/model-factory
+
+Creates a Vercel AI SDK `LanguageModel` from environment variables, connecting to any OpenAI-compatible provider.
+
+---
+
+## Purpose
+
+Without `model-factory`, you would have to hard-code provider setup logic (provider name, base URL, API key, model ID) in every package that needs an LLM. This module centralises that configuration into four environment variables and returns a ready-to-use `LanguageModel` instance.
+
+The design is deliberately minimal: one function, four env vars, no config files. Any OpenAI-compatible API gateway (LiteLLM, OpenRouter, a direct provider endpoint) works without code changes.
+
+## Key Concepts
+
+- **OpenAI-compatible provider** -- The factory uses `@ai-sdk/openai-compatible` under the hood, so any provider that exposes an OpenAI-shaped chat completions API is supported.
+- **Environment-driven configuration** -- All connection details come from `CUSTOM_PROVIDER_*` environment variables. No constructor arguments, no config objects.
+- **Vercel AI SDK integration** -- The returned object satisfies the AI SDK `LanguageModel` interface and can be passed directly to `generateText`, `streamText`, and other AI SDK functions.
+
+---
+
+## Installation
+
+```bash
+pnpm add @auto-engineer/model-factory
+```
+
+Peer dependency:
+
+```bash
+pnpm add ai
+```
+
+---
+
+## Quick Start
+
+Set the required environment variables:
+
+```bash
+export CUSTOM_PROVIDER_NAME="litellm"
+export CUSTOM_PROVIDER_BASE_URL="https://gateway.example.com/v1"
+export CUSTOM_PROVIDER_API_KEY="sk-..."
+export CUSTOM_PROVIDER_DEFAULT_MODEL="xai/grok-code-fast-1"
+```
+
+Then create and use a model:
+
+```typescript
+import { createModelFromEnv } from '@auto-engineer/model-factory';
+import { generateText } from 'ai';
+
+const model = createModelFromEnv();
+
+const { text } = await generateText({
+  model,
+  prompt: 'Explain monads in one sentence.',
+});
+```
+
+---
+
+## How-to Guides
+
+### Connect to a different provider
+
+Change the environment variables -- no code changes required:
+
+```bash
+# OpenRouter
+export CUSTOM_PROVIDER_NAME="openrouter"
+export CUSTOM_PROVIDER_BASE_URL="https://openrouter.ai/api/v1"
+export CUSTOM_PROVIDER_API_KEY="sk-or-..."
+export CUSTOM_PROVIDER_DEFAULT_MODEL="anthropic/claude-sonnet-4-20250514"
+```
+
+### Use in tests
+
+Mock the module or set test-specific env vars before importing:
+
+```typescript
+process.env.CUSTOM_PROVIDER_NAME = 'test-provider';
+process.env.CUSTOM_PROVIDER_BASE_URL = 'http://localhost:4000/v1';
+process.env.CUSTOM_PROVIDER_API_KEY = 'test-key';
+process.env.CUSTOM_PROVIDER_DEFAULT_MODEL = 'test-model';
+
+const { createModelFromEnv } = await import('@auto-engineer/model-factory');
+const model = createModelFromEnv();
+```
+
+---
+
+## API Reference
+
+### Package Exports
+
+| Export | Kind | Description |
+|---|---|---|
+| `createModelFromEnv` | Function | Creates a `LanguageModel` from environment variables |
+| `ChatMessage` | Interface | A single chat message with role and content |
+| `ChatTurn` | Interface | A prompt/response pair |
+
+### Functions
+
+#### `createModelFromEnv()`
+
+Creates a Vercel AI SDK `LanguageModel` by reading four required environment variables.
+
+```typescript
+function createModelFromEnv(): LanguageModel
+```
+
+**Environment variables (all required):**
+
+| Variable | Description |
+|---|---|
+| `CUSTOM_PROVIDER_NAME` | Identifier for the provider (e.g. `"litellm"`) |
+| `CUSTOM_PROVIDER_BASE_URL` | Base URL of the OpenAI-compatible API |
+| `CUSTOM_PROVIDER_API_KEY` | API key for authentication |
+| `CUSTOM_PROVIDER_DEFAULT_MODEL` | Model identifier to use (e.g. `"xai/grok-code-fast-1"`) |
+
+**Throws:** `Error` if any of the four environment variables is missing or empty.
+
+**Returns:** A `LanguageModel` instance backed by the configured provider.
+
+### Interfaces
+
+#### `ChatMessage`
+
+```typescript
+interface ChatMessage {
+  role: 'user' | 'assistant';
+  content: string;
+}
+```
+
+#### `ChatTurn`
+
+```typescript
+interface ChatTurn {
+  prompt: ChatMessage;
+  response: ChatMessage;
+}
+```
+
+---
+
+## Architecture
+
+```
+src/
+  index.ts          # createModelFromEnv, ChatMessage, ChatTurn
+  index.specs.ts    # Unit tests
+```
+
+### Dependencies
+
+| Package | Purpose |
+|---|---|
+| `@ai-sdk/openai-compatible` | Creates the underlying provider client |
+| `ai` (peer) | Provides the `LanguageModel` type |
+
+### Unused direct dependencies
+
+The following packages are listed in `dependencies` but are not currently imported in source code. They may be intended for future provider-specific factories:
+
+| Package |
+|---|
+| `@ai-sdk/anthropic` |
+| `@ai-sdk/google` |
+| `@ai-sdk/openai` |
+| `@ai-sdk/xai` |

package/package.json

@@ -22,7 +22,7 @@
   "publishConfig": {
     "access": "public"
   },
-  "version": "1.148.0",
+  "version": "1.149.0",
   "scripts": {
     "build": "tsc && tsx ../../scripts/fix-esm-imports.ts",
     "test": "vitest run --reporter=dot",