Add OpenAICompat for openai-completions provider quirks

Fixes #133

packages/coding-agent/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+### Added
+
+- **OpenAI compatibility overrides in models.json**: Custom models using `openai-completions` API can now specify a `compat` object to override provider quirks (`supportsStore`, `supportsDeveloperRole`, `supportsReasoningEffort`, `maxTokensField`). Useful for LiteLLM, custom proxies, and other non-standard endpoints. ([#133](https://github.com/badlogic/pi-mono/issues/133), thanks @fink-andreas for the initial idea and PR)
+
 ## [0.13.2] - 2025-12-07
 
 ### Changed

packages/coding-agent/README.md

@@ -315,6 +315,47 @@ You can add custom HTTP headers to bypass Cloudflare bot detection, add authenti
 - **Model-level `headers`**: Additional headers for specific models (merged with provider headers)
 - Model headers override provider headers when keys conflict
 
+### OpenAI Compatibility Settings
+
+The `openai-completions` API is implemented by many providers with minor differences (Ollama, vLLM, LiteLLM, llama.cpp, etc.). By default, compatibility settings are auto-detected from the `baseUrl`. For custom proxies or unknown endpoints, you can override these via the `compat` field on models:
+
+```json
+{
+  "providers": {
+    "litellm": {
+      "baseUrl": "http://localhost:4000/v1",
+      "apiKey": "LITELLM_API_KEY",
+      "api": "openai-completions",
+      "models": [
+        {
+          "id": "gpt-4o",
+          "name": "GPT-4o (via LiteLLM)",
+          "reasoning": false,
+          "input": ["text", "image"],
+          "cost": {"input": 2.5, "output": 10, "cacheRead": 0, "cacheWrite": 0},
+          "contextWindow": 128000,
+          "maxTokens": 16384,
+          "compat": {
+            "supportsStore": false
+          }
+        }
+      ]
+    }
+  }
+}
+```
+
+Available `compat` fields (all optional, auto-detected if not set):
+
+| Field | Type | Default | Description |
+|-------|------|---------|-------------|
+| `supportsStore` | boolean | auto | Whether provider supports the `store` field |
+| `supportsDeveloperRole` | boolean | auto | Whether provider supports `developer` role (vs `system`) |
+| `supportsReasoningEffort` | boolean | auto | Whether provider supports `reasoning_effort` parameter |
+| `maxTokensField` | string | auto | Use `"max_completion_tokens"` or `"max_tokens"` |
+
+If `compat` is partially set, unspecified fields use auto-detected values.
+
 ### Authorization Header
 
 Some providers require an explicit `Authorization: Bearer <token>` header. Set `authHeader: true` to automatically add this header using the resolved `apiKey`:

packages/coding-agent/src/model-config.ts

@@ -9,6 +9,14 @@ import { loadOAuthCredentials } from "./oauth/storage.js";
 // Handle both default and named exports
 const Ajv = (AjvModule as any).default || AjvModule;
 
+// Schema for OpenAI compatibility settings
+const OpenAICompatSchema = Type.Object({
+	supportsStore: Type.Optional(Type.Boolean()),
+	supportsDeveloperRole: Type.Optional(Type.Boolean()),
+	supportsReasoningEffort: Type.Optional(Type.Boolean()),
+	maxTokensField: Type.Optional(Type.Union([Type.Literal("max_completion_tokens"), Type.Literal("max_tokens")])),
+});
+
 // Schema for custom model definition
 const ModelDefinitionSchema = Type.Object({
 	id: Type.String({ minLength: 1 }),
@@ -32,6 +40,7 @@ const ModelDefinitionSchema = Type.Object({
 	contextWindow: Type.Number(),
 	maxTokens: Type.Number(),
 	headers: Type.Optional(Type.Record(Type.String(), Type.String())),
+	compat: Type.Optional(OpenAICompatSchema),
 });
 
 const ProviderConfigSchema = Type.Object({
@@ -201,7 +210,8 @@ function parseModels(config: ModelsConfig): Model<Api>[] {
 				contextWindow: modelDef.contextWindow,
 				maxTokens: modelDef.maxTokens,
 				headers,
-			});
+				compat: modelDef.compat,
+			} as Model<Api>);
 		}
 	}