deepagents-code 0.1.14__tar.gz → 0.1.16__tar.gz

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/CHANGELOG.md

@@ -2,6 +2,19 @@
 
 # Deep Agents Code Changelog
 
+## [0.1.16](https://github.com/langchain-ai/deepagents/compare/deepagents-code==0.1.15...deepagents-code==0.1.16) (2026-06-13)
+
+### Features
+
+* ChatGPT OAuth sign-in for Codex models ([#3532](https://github.com/langchain-ai/deepagents/issues/3532)) ([202e0bd](https://github.com/langchain-ai/deepagents/commit/202e0bd3e5b8b874a69656815489308d75a77d05))
+* Add Vercel Sandbox provider ([#3588](https://github.com/langchain-ai/deepagents/issues/3588)) ([e5e4748](https://github.com/langchain-ai/deepagents/commit/e5e4748cb6c66ddaa9444ab464990c1a5d10854d))
+
+## [0.1.15](https://github.com/langchain-ai/deepagents/compare/deepagents-code==0.1.14...deepagents-code==0.1.15) (2026-06-12)
+
+### Features
+
+* Add Deep Agents client version metadata ([#3934](https://github.com/langchain-ai/deepagents/issues/3934)) ([058cec6](https://github.com/langchain-ai/deepagents/commit/058cec654645dd1a504408d712ab678edc48273d))
+
 ## [0.1.14](https://github.com/langchain-ai/deepagents/compare/deepagents-code==0.1.13...deepagents-code==0.1.14) (2026-06-12)
 
 ### Features

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepagents-code
-Version: 0.1.14
+Version: 0.1.16
 Summary: Terminal interface for Deep Agents - interactive AI agent with file operations, shell access, and sub-agent capabilities.
 Project-URL: Homepage, https://docs.langchain.com/oss/python/deepagents/overview
 Project-URL: Documentation, https://reference.langchain.com/python/deepagents/
@@ -27,13 +27,13 @@ Classifier: Topic :: Terminals
 Requires-Python: <4.0,>=3.11
 Requires-Dist: aiosqlite<1.0.0,>=0.22.1
 Requires-Dist: deepagents-acp<1.0.0,>=0.0.8
-Requires-Dist: deepagents==0.6.8
+Requires-Dist: deepagents==0.6.10
 Requires-Dist: httpx<1.0.0,>=0.28.1
-Requires-Dist: langchain-anthropic<2.0.0,>=1.4.5
+Requires-Dist: langchain-anthropic<2.0.0,>=1.4.6
 Requires-Dist: langchain-google-genai<5.0.0,>=4.2.5
 Requires-Dist: langchain-mcp-adapters<1.0.0,>=0.3.0
-Requires-Dist: langchain-openai<2.0.0,>=1.3.0
-Requires-Dist: langchain<2.0.0,>=1.3.8
+Requires-Dist: langchain-openai<2.0.0,>=1.3.2
+Requires-Dist: langchain<2.0.0,>=1.3.9
 Requires-Dist: langgraph-checkpoint-sqlite<4.0.0,>=3.1.0
 Requires-Dist: langgraph-cli[inmem]<1.0.0,>=0.4.29
 Requires-Dist: langgraph-runtime-inmem<1.0.0,>=0.30.0
@@ -57,7 +57,7 @@ Requires-Dist: uuid-utils<1.0.0,>=0.16.0
 Provides-Extra: agentcore
 Requires-Dist: langchain-agentcore-codeinterpreter<1.0.0,>=0.0.4; extra == 'agentcore'
 Provides-Extra: all-providers
-Requires-Dist: langchain-anthropic<2.0.0,>=1.4.5; extra == 'all-providers'
+Requires-Dist: langchain-anthropic<2.0.0,>=1.4.6; extra == 'all-providers'
 Requires-Dist: langchain-aws<2.0.0,>=1.5.1; extra == 'all-providers'
 Requires-Dist: langchain-baseten<1.0.0,>=0.2.0; extra == 'all-providers'
 Requires-Dist: langchain-cohere<1.0.0,>=0.6.0; extra == 'all-providers'
@@ -72,7 +72,7 @@ Requires-Dist: langchain-litellm<2.0.0,>=0.6.6; extra == 'all-providers'
 Requires-Dist: langchain-mistralai<2.0.0,>=1.1.5; extra == 'all-providers'
 Requires-Dist: langchain-nvidia-ai-endpoints<2.0.0,>=1.4.1; extra == 'all-providers'
 Requires-Dist: langchain-ollama<2.0.0,>=1.1.0; extra == 'all-providers'
-Requires-Dist: langchain-openai<2.0.0,>=1.3.0; extra == 'all-providers'
+Requires-Dist: langchain-openai<2.0.0,>=1.3.2; extra == 'all-providers'
 Requires-Dist: langchain-openrouter<2.0.0,>=0.2.3; extra == 'all-providers'
 Requires-Dist: langchain-perplexity<2.0.0,>=1.4.0; extra == 'all-providers'
 Requires-Dist: langchain-together<2.0.0,>=0.4.0; extra == 'all-providers'
@@ -82,8 +82,9 @@ Requires-Dist: langchain-agentcore-codeinterpreter<1.0.0,>=0.0.4; extra == 'all-
 Requires-Dist: langchain-daytona>=0.0.7; extra == 'all-sandboxes'
 Requires-Dist: langchain-modal>=0.0.5; extra == 'all-sandboxes'
 Requires-Dist: langchain-runloop>=0.0.6; extra == 'all-sandboxes'
+Requires-Dist: langchain-vercel-sandbox>=0.0.1; extra == 'all-sandboxes'
 Provides-Extra: anthropic
-Requires-Dist: langchain-anthropic<2.0.0,>=1.4.5; extra == 'anthropic'
+Requires-Dist: langchain-anthropic<2.0.0,>=1.4.6; extra == 'anthropic'
 Provides-Extra: baseten
 Requires-Dist: langchain-baseten<1.0.0,>=0.2.0; extra == 'baseten'
 Provides-Extra: bedrock
@@ -115,7 +116,7 @@ Requires-Dist: langchain-nvidia-ai-endpoints<2.0.0,>=1.4.1; extra == 'nvidia'
 Provides-Extra: ollama
 Requires-Dist: langchain-ollama<2.0.0,>=1.1.0; extra == 'ollama'
 Provides-Extra: openai
-Requires-Dist: langchain-openai<2.0.0,>=1.3.0; extra == 'openai'
+Requires-Dist: langchain-openai<2.0.0,>=1.3.2; extra == 'openai'
 Provides-Extra: openrouter
 Requires-Dist: langchain-openrouter<2.0.0,>=0.2.3; extra == 'openrouter'
 Provides-Extra: perplexity
@@ -126,6 +127,8 @@ Provides-Extra: runloop
 Requires-Dist: langchain-runloop>=0.0.6; extra == 'runloop'
 Provides-Extra: together
 Requires-Dist: langchain-together<2.0.0,>=0.4.0; extra == 'together'
+Provides-Extra: vercel
+Requires-Dist: langchain-vercel-sandbox>=0.0.1; extra == 'vercel'
 Provides-Extra: vertex
 Requires-Dist: langchain-google-vertexai<4.0.0,>=3.2.4; extra == 'vertex'
 Provides-Extra: xai
@@ -182,6 +185,8 @@ The fastest way to start using Deep Agents. `deepagents-code` is a pre-built cod
 - **[Changelog](https://github.com/langchain-ai/deepagents/blob/main/libs/code/CHANGELOG.md)**
 - **[Source code](https://github.com/langchain-ai/deepagents/tree/main/libs/code)**
 - **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — underlying agent harness
+- [LangChain Academy](https://academy.langchain.com/) — Comprehensive, free courses on LangChain libraries and products, made by the LangChain team.
+- [Code of Conduct](https://github.com/langchain-ai/langchain/?tab=coc-ov-file) — community guidelines and standards
 
 ## 📕 Releases & Versioning
 

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/README.md

@@ -48,6 +48,8 @@ The fastest way to start using Deep Agents. `deepagents-code` is a pre-built cod
 - **[Changelog](https://github.com/langchain-ai/deepagents/blob/main/libs/code/CHANGELOG.md)**
 - **[Source code](https://github.com/langchain-ai/deepagents/tree/main/libs/code)**
 - **[Deep Agents SDK](https://github.com/langchain-ai/deepagents)** — underlying agent harness
+- [LangChain Academy](https://academy.langchain.com/) — Comprehensive, free courses on LangChain libraries and products, made by the LangChain team.
+- [Code of Conduct](https://github.com/langchain-ai/langchain/?tab=coc-ov-file) — community guidelines and standards
 
 ## 📕 Releases & Versioning
 

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/_cli_context.py

@@ -7,9 +7,17 @@ middleware stack.
 
 from __future__ import annotations
 
-from typing import Any
+from dataclasses import dataclass, field
+from typing import Any, TypedDict
 
-from typing_extensions import TypedDict
+
+@dataclass
+class CLIContextSchema:
+    """Server-side schema for run-scoped CLI context."""
+
+    model: str | None = None
+    model_params: dict[str, Any] = field(default_factory=dict)
+    effective_model: str | None = None
 
 
 class CLIContext(TypedDict, total=False):

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/_env_vars.py

@@ -170,7 +170,6 @@ THEME = "DEEPAGENTS_CODE_THEME"
 USER_ID = "DEEPAGENTS_CODE_USER_ID"
 """Attach a user identifier to LangSmith trace metadata."""
 
-
 _TRUTHY_VALUES = frozenset({"1", "true", "yes", "on"})
 _FALSY_VALUES = frozenset({"0", "false", "no", "off", ""})
 

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/_version.py

@@ -2,7 +2,7 @@
 
 # Keep the `x-release-please-version` annotation — release-please uses it to
 # bump `__version__` in sync with `pyproject.toml` on every release PR.
-__version__ = "0.1.14"  # x-release-please-version
+__version__ = "0.1.16"  # x-release-please-version
 
 DOCS_URL = "https://docs.langchain.com/oss/python/deepagents/code"
 """URL for `deepagents-code` documentation."""

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/agent.py

@@ -52,6 +52,7 @@ if TYPE_CHECKING:
 from langchain.agents.middleware.types import AgentMiddleware
 
 from deepagents_code import theme
+from deepagents_code._cli_context import CLIContextSchema
 from deepagents_code._constants import DEFAULT_AGENT_NAME
 from deepagents_code.config import (
     _INHERITED_PYTHONPATH_ENV,
@@ -1118,7 +1119,7 @@ def create_cli_agent(
     cwd: str | Path | None = None,
     project_context: ProjectContext | None = None,
     async_subagents: list[AsyncSubAgent] | None = None,
-) -> tuple[Pregel, CompositeBackend]:
+) -> tuple[Pregel[Any, Any, Any, Any], CompositeBackend]:
     """Create a CLI-configured agent with flexible options.
 
     This is the main entry point for creating a Deep Agents Code agent, usable
@@ -1578,6 +1579,7 @@ def create_cli_agent(
         backend=composite_backend,
         middleware=agent_middleware,
         interrupt_on=interrupt_on,
+        context_schema=CLIContextSchema,
         checkpointer=checkpointer,
         subagents=all_subagents or None,
         name=_sanitize_agent_message_name(assistant_id),

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/auth_display.py

@@ -7,6 +7,7 @@ from typing import TYPE_CHECKING, assert_never
 from textual.content import Content
 
 from deepagents_code.model_config import (
+    CODEX_PROVIDER,
     ProviderAuthSource,
     ProviderAuthState,
     ProviderAuthStatus,
@@ -29,6 +30,13 @@ def format_auth_badge(status: ProviderAuthStatus) -> Content:
     Returns:
         A styled badge `Content` for the auth manager surface.
     """
+    # The ChatGPT-OAuth codex provider has no API key, so its credential is a
+    # file-backed OAuth token rather than a stored key. Give it distinctive
+    # badges (`[chatgpt]` / `[sign in to chatgpt]`) so users don't read the
+    # generic `[stored]`/`[missing]` as a literal API key on disk.
+    if status.provider == CODEX_PROVIDER:
+        return _format_codex_badge(status)
+
     state = status.state
     match state:
         case ProviderAuthState.CONFIGURED:
@@ -103,6 +111,48 @@ def _auth_badge(detail: str, *, prefix: str = "") -> Content:
     )
 
 
+def _format_codex_badge(status: ProviderAuthStatus) -> Content:
+    """Format the auth manager badge for the ChatGPT-OAuth codex provider.
+
+    Codex signs in through ChatGPT OAuth rather than an API key, so a
+    `CONFIGURED` status renders as `[chatgpt]` (carrying the plan name when the
+    status detail includes one) and any other state renders as a
+    `[sign in to chatgpt]` prompt.
+
+    Args:
+        status: The codex provider's auth status.
+
+    Returns:
+        A styled badge reflecting ChatGPT sign-in state.
+    """
+    if status.state is ProviderAuthState.CONFIGURED:
+        badge_text = "[chatgpt]"
+        # `_get_codex_auth_status` encodes the plan as a trailing "(plan)" in
+        # the detail string (e.g. "signed in to ChatGPT (pro)").
+        if status.detail and (plan := _codex_plan_from_detail(status.detail)):
+            badge_text = f"[chatgpt: {plan}]"
+        return Content.styled(badge_text, "bold $success")
+    return Content.styled("[sign in to chatgpt]", "bold $warning")
+
+
+def _codex_plan_from_detail(detail: str) -> str | None:
+    """Extract the ChatGPT plan from a codex auth detail string.
+
+    Args:
+        detail: Human-readable codex auth status detail.
+
+    Returns:
+        The ChatGPT plan text, or `None` when no bounded plan is present.
+    """
+    _, marker, plan_tail = detail.partition("(")
+    if not marker:
+        return None
+    plan, marker, _ = plan_tail.partition(")")
+    if not marker or not plan:
+        return None
+    return plan
+
+
 def _format_configured_badge(status: ProviderAuthStatus) -> Content:
     """Format the auth manager badge for a `CONFIGURED` provider.
 

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/config.py

@@ -13,7 +13,7 @@ import sys
 import threading
 from dataclasses import dataclass
 from enum import StrEnum
-from importlib.metadata import PackageNotFoundError, distribution
+from importlib.metadata import PackageNotFoundError, distribution, version
 from pathlib import Path
 from typing import TYPE_CHECKING, Any
 from urllib.parse import unquote, urlparse
@@ -568,6 +568,23 @@ Kept short so tracing metadata can never stall app flows.
 """
 
 
+def _get_deepagents_version() -> str | None:
+    """Read the installed Deep Agents SDK version from package metadata.
+
+    Returns:
+        The installed Deep Agents SDK version, or `None` when package metadata
+        is unavailable.
+    """
+    try:
+        return version("deepagents")
+    except PackageNotFoundError:
+        logger.debug(
+            "Failed to read deepagents version from package metadata",
+            exc_info=True,
+        )
+        return None
+
+
 def _resolve_editable_info() -> tuple[bool, str | None]:
     """Parse PEP 610 `direct_url.json` once and cache both results.
 
@@ -826,11 +843,15 @@ def build_stream_config(
 ) -> RunnableConfig:
     """Build the LangGraph stream config dict.
 
-    Injects the dcode version into `metadata["versions"]` so LangSmith traces
+    Injects the dcode version into `metadata["lc_versions"]` so LangSmith traces
     can be correlated with specific releases. `create_deep_agent` supplies the
     SDK version through the compiled graph config, and LangChain merges nested
     metadata dictionaries so both versions survive at stream time.
 
+    Also records `dcode_client_deepagents_version` as a dcode-client diagnostic.
+    This describes the Deep Agents package installed alongside the TUI, which
+    can differ from a remote graph's Deep Agents runtime version.
+
     Includes `ls_integration` metadata so LangSmith traces originating from
     the app are distinguishable from bare SDK usage.
 
@@ -854,9 +875,13 @@ def build_stream_config(
         cwd = ""
 
     metadata: dict[str, Any] = {
-        "versions": {"deepagents-code": __version__},
+        "lc_versions": {"deepagents-code": __version__},
         "ls_integration": "deepagents-code",
     }
+    deepagents_version = _get_deepagents_version()
+    if deepagents_version is not None:
+        metadata["dcode_client_deepagents_version"] = deepagents_version
+
     from deepagents_code._env_vars import USER_ID
 
     user_id = os.environ.get(USER_ID)
@@ -3045,6 +3070,8 @@ def create_model(
         warn_on_split_credential_source(provider)
         apply_stored_credentials(provider)
 
+    from deepagents_code.model_config import CODEX_PROVIDER
+
     # Early credential check — fail fast with an actionable message instead of
     # letting the provider SDK raise an opaque auth error on first invocation.
     # Providers that support implicit auth (e.g., Vertex AI ADC) are excluded
@@ -3054,6 +3081,13 @@ def create_model(
         if cred_status is False:
             from deepagents_code.model_config import MissingCredentialsError
 
+            if provider == CODEX_PROVIDER:
+                # No env var to set; point the user at `/auth` instead.
+                msg = (
+                    "Not signed in to ChatGPT. Run `/auth` and select "
+                    "openai_codex to sign in with your ChatGPT account."
+                )
+                raise MissingCredentialsError(msg, provider=provider, env_var=None)
             env_var = get_credential_env_var(provider)
             display_env = env_var or f"<{provider} API key>"
             msg = (
@@ -3117,7 +3151,45 @@ def create_model(
     config = ModelConfig.load()
     class_path = config.get_class_path(provider) if provider else None
 
-    if class_path:
+    if provider == CODEX_PROVIDER:
+        # Codex models are constructed directly via `_ChatOpenAICodex` so the
+        # `token_provider=` kwarg is wired to the on-disk OAuth token store
+        # before any request goes out. `init_chat_model` does not know about
+        # this class and would route through API-key `ChatOpenAI` instead.
+        from deepagents_code.integrations import openai_codex as _codex
+        from deepagents_code.model_config import (
+            MissingCredentialsError,
+            ModelConfigError,
+        )
+
+        # Drop any `api_key` left in kwargs (e.g. from a config-level
+        # `api_key_env` set on the codex provider, or a `--model-params
+        # api_key=...`) so the bearer always comes from the OAuth
+        # `token_provider` rather than a static key.
+        kwargs.pop("api_key", None)
+        try:
+            model = _codex.build_chat_model(model_name, **kwargs)
+        except FileNotFoundError as exc:
+            msg = (
+                "Not signed in to ChatGPT. Run `/auth` and select "
+                "openai_codex to sign in with your ChatGPT account."
+            )
+            raise MissingCredentialsError(msg, provider=provider, env_var=None) from exc
+        except _codex.CodexAuthExpiredError as exc:
+            # A token exists but its refresh token is dead. Route through the
+            # same `MissingCredentialsError` recovery path as a missing token
+            # (which the retry flow re-attempts after `/auth`) instead of the
+            # generic `ModelConfigError` below, which would not offer sign-in.
+            msg = (
+                "ChatGPT session expired. Run `/auth` and select openai_codex "
+                "to sign in again."
+            )
+            raise MissingCredentialsError(msg, provider=provider, env_var=None) from exc
+        except Exception as exc:
+            spec = f"{provider}:{model_name}"
+            msg = f"Failed to initialize Codex model '{spec}': {exc}"
+            raise ModelConfigError(msg) from exc
+    elif class_path:
         model = _create_model_from_class(class_path, model_name, provider, kwargs)
     else:
         model = _create_model_via_init(model_name, provider, kwargs)

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/configurable_model.py

@@ -7,6 +7,7 @@ the graph.
 
 from __future__ import annotations
 
+import asyncio
 import logging
 from typing import TYPE_CHECKING, Any
 
@@ -17,9 +18,13 @@ from langchain.agents.middleware.types import (
     ModelResponse,
 )
 
+from deepagents_code._cli_context import CLIContextSchema
+
 if TYPE_CHECKING:
     from collections.abc import Awaitable, Callable
 
+    from deepagents_code.config import ModelResult
+
 
 logger = logging.getLogger(__name__)
 
@@ -56,55 +61,54 @@ _ANTHROPIC_ONLY_SETTINGS: set[str] = {"cache_control"}
 must be stripped on cross-provider swap."""
 
 
-def _apply_overrides(request: ModelRequest) -> ModelRequest:
-    """Apply model/param overrides from `CLIContext` on the runtime.
+def _get_context(request: ModelRequest) -> CLIContextSchema | None:
+    """Return runtime context when it matches the CLI context shape."""
+    runtime = request.runtime
+    if runtime is None:
+        return None
 
-    Reads `'model'` and `'model_params'` from `runtime.context` and, when
-    present, swaps the model and/or merges extra settings into the request.
-    On a cross-provider swap away from Anthropic, Anthropic-only settings
-    (e.g. `cache_control`) are stripped. The `### Model Identity` section
-    in the system prompt is also patched to reflect the new model.
+    ctx = runtime.context
+    if isinstance(ctx, CLIContextSchema):
+        return ctx
+    if isinstance(ctx, dict):
+        return CLIContextSchema(
+            model=ctx.get("model"),
+            model_params=ctx.get("model_params") or {},
+            effective_model=ctx.get("effective_model"),
+        )
+    return None
+
+
+def _build_overrides(
+    request: ModelRequest, ctx: CLIContextSchema, model_result: ModelResult | None
+) -> ModelRequest:
+    """Build the overridden request from a (possibly resolved) model result.
+
+    Holds the post-construction logic shared by the sync and async override
+    paths: applying the model swap, merging `model_params`, stripping
+    Anthropic-only settings on a cross-provider swap, and patching the
+    `### Model Identity` system-prompt section. The only thing that differs
+    between the two callers is how `model_result` is produced (a direct
+    `create_model` call vs. an `asyncio.to_thread` offload).
 
     Args:
         request: The incoming model request from the middleware chain.
+        ctx: Runtime CLI context carrying the requested overrides.
+        model_result: The resolved model result from `create_model`, or `None`
+            when no model swap was requested.
 
     Returns:
-        The original request unchanged when no `CLIContext` is present or it
-            contains no overrides, otherwise a new request with overrides
-            applied via `request.override()`.
+        The original request when no overrides apply, otherwise a new request
+            with overrides applied via `request.override()`.
     """
-    runtime = request.runtime
-    if runtime is None:
-        return request
-
-    ctx = runtime.context
-    if not isinstance(ctx, dict):
-        return request
-
     overrides: dict[str, Any] = {}
 
-    # Model swap
-    new_model = None
-    model = ctx.get("model")
-    if model and not model_matches_spec(request.model, model):
-        from deepagents_code.config import create_model
-        from deepagents_code.model_config import ModelConfigError
-
-        logger.debug("Overriding model to %s", model)
-        try:
-            model_result = create_model(model)
-            new_model = model_result.model
-        except ModelConfigError:
-            logger.exception(
-                "Failed to resolve runtime model override '%s'; "
-                "continuing with current model",
-                model,
-            )
-            return request
+    new_model = model_result.model if model_result is not None else None
+    if new_model is not None:
         overrides["model"] = new_model
 
     # Param merge
-    model_params = ctx.get("model_params", {})
+    model_params = ctx.model_params
     if model_params:
         overrides["model_settings"] = {**request.model_settings, **model_params}
 
@@ -131,7 +135,7 @@ def _apply_overrides(request: ModelRequest) -> ModelRequest:
     # We read metadata from model_result (not the app's settings singleton)
     # because the middleware runs in the server subprocess where settings
     # are never updated by /model.
-    if new_model is not None and request.system_prompt:
+    if model_result is not None and request.system_prompt:
         from deepagents_code.agent import (
             MODEL_IDENTITY_RE,
             build_model_identity_section,
@@ -159,6 +163,78 @@ def _apply_overrides(request: ModelRequest) -> ModelRequest:
     return request.override(**overrides)
 
 
+def _apply_overrides(request: ModelRequest) -> ModelRequest:
+    """Apply model/param overrides from `CLIContext` on the runtime.
+
+    Reads `'model'` and `'model_params'` from `runtime.context` and, when
+    present, swaps the model and/or merges extra settings into the request.
+    On a cross-provider swap away from Anthropic, Anthropic-only settings
+    (e.g. `cache_control`) are stripped. The `### Model Identity` section
+    in the system prompt is also patched to reflect the new model.
+
+    Args:
+        request: The incoming model request from the middleware chain.
+
+    Returns:
+        The original request unchanged when no `CLIContext` is present or it
+            contains no overrides, otherwise a new request with overrides
+            applied via `request.override()`.
+    """
+    ctx = _get_context(request)
+    if ctx is None:
+        return request
+
+    model_result = None
+    model = ctx.model
+    if model and not model_matches_spec(request.model, model):
+        from deepagents_code.config import create_model
+        from deepagents_code.model_config import ModelConfigError
+
+        logger.debug("Overriding model to %s", model)
+        try:
+            model_result = create_model(model)
+        except ModelConfigError:
+            logger.exception(
+                "Failed to resolve runtime model override '%s'; "
+                "continuing with current model",
+                model,
+            )
+            return request
+
+    return _build_overrides(request, ctx, model_result)
+
+
+async def _apply_overrides_async(request: ModelRequest) -> ModelRequest:
+    """Async variant of `_apply_overrides` that offloads model construction.
+
+    Returns:
+        The original request when no async override applies, otherwise a request
+            with the runtime model or settings override applied.
+    """
+    ctx = _get_context(request)
+    if ctx is None:
+        return request
+
+    model_result = None
+    model = ctx.model
+    if model and not model_matches_spec(request.model, model):
+        from deepagents_code.config import create_model
+        from deepagents_code.model_config import ModelConfigError
+
+        logger.debug("Overriding model to %s", model)
+        try:
+            model_result = await asyncio.to_thread(create_model, model)
+        except ModelConfigError:
+            logger.exception(
+                "Failed to resolve runtime model override '%s'; "
+                "continuing with current model",
+                model,
+            )
+            return request
+
+    return _build_overrides(request, ctx, model_result)
+
+
 class ConfigurableModelMiddleware(AgentMiddleware):
     """Swap the model or per-call settings from `runtime.context`.
 
@@ -198,4 +274,4 @@ class ConfigurableModelMiddleware(AgentMiddleware):
         Returns:
             The `ModelResponse` produced by the downstream handler.
         """
-        return await handler(_apply_overrides(request))
+        return await handler(await _apply_overrides_async(request))

{deepagents_code-0.1.14 → deepagents_code-0.1.16}/deepagents_code/extras_info.py

@@ -66,7 +66,9 @@ MODEL_PROVIDER_EXTRAS: frozenset[str] = frozenset(
 Keep in sync with `[project.optional-dependencies]` in `pyproject.toml`.
 """
 
-SANDBOX_EXTRAS: frozenset[str] = frozenset({"agentcore", "daytona", "modal", "runloop"})
+SANDBOX_EXTRAS: frozenset[str] = frozenset(
+    {"agentcore", "daytona", "modal", "runloop", "vercel"}
+)
 """Optional extras that add sandbox integrations."""
 
 STANDALONE_EXTRAS: frozenset[str] = frozenset({"quickjs"})