feat(BA-5528): add deployment chat CLI for vLLM-backed model services

changes/11344.feature.md

@@ -0,0 +1 @@
+Add `./bai deployment chat` for one-shot OpenAI-compatible chat against deployed inference services.

src/ai/backend/client/cli/v2/deployment/__init__.py

@@ -1,5 +1,6 @@
 from .access_token import access_token
 from .auto_scaling_rule import auto_scaling_rule
+from .chat import chat, chat_cache, chat_config, chat_history
 from .commands import deployment as deployment
 from .options import options
 from .policy import policy
@@ -15,5 +16,9 @@
 deployment.add_command(access_token)
 deployment.add_command(auto_scaling_rule)
 deployment.add_command(options)
+deployment.add_command(chat)
+deployment.add_command(chat_config)
+deployment.add_command(chat_cache)
+deployment.add_command(chat_history)
 
 __all__ = ("deployment",)

src/ai/backend/client/cli/v2/deployment/chat/__init__.py

@@ -0,0 +1,17 @@
+"""``./bai deployment chat``, ``chat-config``, ``chat-cache``, and ``chat-history`` CLI commands.
+
+Submodules:
+- :mod:`commands` — Click command/group definitions.
+- :mod:`types` — Pydantic models for the on-disk cache, config, and history,
+  including the ``.load()``/``.save()`` classmethods that wire them to
+  ``~/.backend.ai/deployment_chat/*.json``.
+- :mod:`utils` — file paths and shared JSON I/O helpers.
+- :mod:`formatter` — display helpers (``mask_token``, ``DeploymentChatFormatter``).
+
+OpenAI-compat wire DTOs live in :mod:`ai.backend.common.dto.clients.openai_compat`
+so they can be reused by any backend.ai component.
+"""
+
+from .commands import chat, chat_cache, chat_config, chat_history
+
+__all__ = ("chat", "chat_cache", "chat_config", "chat_history")