docs: fix openai-py base_url examples (must include /v1)

live testing surfaced — openai-py library appends /chat/completions to
base_url, so base_url must include /v1. previously docs showed:
  base_url="https://hub.example.com/zai"
which produces a 404 since the hub serves /zai/v1/chat/completions.

correct usage shown now in:
- README.md (added a 'Calling from the OpenAI Python library' subsection
  with both non-streaming and streaming examples)
- examples/ZAI_PUBLISH.md (corrected and called out the /v1 requirement
  explicitly)

verified live: openai-py 1.x against zhub hub, both stream=False and
stream=True paths work cleanly when /v1 is in base_url. signed manifest
fetch + external verify also confirmed working — verify_manifest()
returns True against /<name>/manifest.json fetched over HTTP.

key pinning verified live: stolen api_key + new ed25519 keypair →
hub returns 'register_failed: key pinning: stored public_key does not
match'. attacker with just an api_key cannot take over a signed
publisher.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: Zawwarsami16

README.md

@@ -60,6 +60,25 @@ python examples/orchestrate_demo.py
 
 The AI saw a connection arrive, registered its capabilities, and invoked them at the user's request — all through the hub, in a single conversation.
 
+### Calling from the OpenAI Python library
+
+```python
+from openai import OpenAI
+
+# IMPORTANT: openai-py expects /v1 in the base_url.
+# Use ".../<name>/v1", not just ".../<name>".
+client = OpenAI(
+    base_url="http://localhost:8080/echo/v1",
+    api_key="zk_...",
+)
+client.chat.completions.create(model="echo", messages=[{"role":"user","content":"hi"}])
+# streaming works too:
+for chunk in client.chat.completions.create(
+    model="echo", messages=[...], stream=True,
+):
+    print(chunk.choices[0].delta.content or "", end="", flush=True)
+```
+
 ---
 
 ## Install

examples/ZAI_PUBLISH.md

@@ -70,8 +70,9 @@ curl https://<hub>.trycloudflare.com/zai/v1/chat/completions \
 ```python
 from openai import OpenAI
 
+# IMPORTANT: openai-py expects /v1 in the base_url. Use ".../zai/v1", not ".../zai".
 client = OpenAI(
-    base_url="https://<hub>.trycloudflare.com/zai",
+    base_url="https://<hub>.trycloudflare.com/zai/v1",
     api_key="zk_a8f2c9d3...",
 )