feat: add default inference endpoint

README.md

@@ -6,7 +6,7 @@ colorTo: gray
 sdk: docker
 pinned: false
 app_port: 8000
-base_path: /web
+base_path: /
 tags:
   - openenv
   - finance
@@ -145,6 +145,7 @@ uvx --from openenv-core openenv validate --url http://localhost:8000
 The root [inference.py](./inference.py) script is the reproducible baseline.
 
 - OpenAI Python client
+- default `ENV_URL`: `https://ehsaaniqbal-invoiceops-env.hf.space`
 - default `API_BASE_URL`: `https://router.huggingface.co/v1`
 - default `MODEL_NAME`: `zai-org/GLM-5.1`
 - fallback tasks are zeroed in `mean_score` by default while raw environment scores are still preserved
@@ -161,10 +162,11 @@ Run it with:
 ```bash
 cd invoiceops_env
 HF_TOKEN=... \
-API_BASE_URL=https://router.huggingface.co/v1 \
 uv run python inference.py
 ```
 
+This matches the competition-style run: `inference.py` talks to the deployed Space by default and uses the Hugging Face router unless you override `ENV_URL`, `API_BASE_URL`, or `MODEL_NAME`.
+
 Optional environment variables:
 
 - `HF_TOKEN`

inference.py

@@ -29,7 +29,7 @@ from invoiceops_env.models import (
     TaskId,
 )
 
-ENV_URL = os.getenv("ENV_URL", "http://localhost:8000")
+ENV_URL = os.getenv("ENV_URL", "https://ehsaaniqbal-invoiceops-env.hf.space")
 DEFAULT_HF_ROUTER_BASE_URL = "https://router.huggingface.co/v1"
 API_BASE_URL = os.getenv("API_BASE_URL", DEFAULT_HF_ROUTER_BASE_URL)
 MODEL_NAME = os.getenv("MODEL_NAME", "zai-org/GLM-5.1")