azurefunctions-agents-extensions-agent-framework 1.0.0b1__tar.gz

azurefunctions_agents_extensions_agent_framework-1.0.0b1/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) Microsoft Corporation.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

azurefunctions_agents_extensions_agent_framework-1.0.0b1/MANIFEST.in

@@ -0,0 +1,3 @@
+recursive-include azurefunctions *.py *.pyi
+recursive-include tests *.py
+include LICENSE README.md

azurefunctions_agents_extensions_agent_framework-1.0.0b1/PKG-INFO

@@ -0,0 +1,195 @@
+Metadata-Version: 2.4
+Name: azurefunctions-agents-extensions-agent-framework
+Version: 1.0.0b1
+Summary: Microsoft Agent Framework integration for Azure Functions.
+Author-email: "Azure Functions team at Microsoft Corp." <azurefunctions@microsoft.com>
+License: MIT License
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Intended Audience :: Developers
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Classifier: Operating System :: Microsoft :: Windows
+Classifier: Operating System :: POSIX
+Classifier: Operating System :: MacOS :: MacOS X
+Classifier: Environment :: Web Environment
+Classifier: Development Status :: 3 - Alpha
+Requires-Python: >=3.13
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: agent-framework-core<2,>=1.13.0
+Requires-Dist: azurefunctions-agents-extensions-base>=1.0.0b1
+Provides-Extra: mcp
+Requires-Dist: azure-identity<2,>=1.25.3; extra == "mcp"
+Requires-Dist: httpx<1,>=0.27; extra == "mcp"
+Requires-Dist: mcp<2,>=1.28.1; extra == "mcp"
+Provides-Extra: durable
+Requires-Dist: azurefunctions-agents-extensions-base[durable]>=1.0.0b1; extra == "durable"
+Provides-Extra: dev
+Requires-Dist: azure-functions-durable>=2.0.0b2; extra == "dev"
+Requires-Dist: coverage; extra == "dev"
+Requires-Dist: flake8; extra == "dev"
+Requires-Dist: mypy; extra == "dev"
+Requires-Dist: pre-commit; extra == "dev"
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: pytest-cov; extra == "dev"
+Requires-Dist: pytest-instafail; extra == "dev"
+Dynamic: license-file
+
+# Azure Functions Microsoft Agent Framework Extension
+
+Inject Microsoft Agent Framework Agents built from raw `.agent.md` instructions
+into Python Azure Functions.
+
+## Install
+
+```text
+pip install azurefunctions-agents-extensions-agent-framework
+```
+
+Install Durable Functions support with the durable extra:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[durable]"
+```
+
+
+Install remote MCP transport and Entra support with the MCP extra:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[mcp]"
+```
+
+## Use an Agent app
+
+Create a zero-argument factory that returns a fresh MAF chat client. A new
+client and Agent context are created and closed for every Function invocation.
+
+```python
+import azure.functions as func
+from agent_framework import Agent
+from azurefunctions.agents.extensions.agent_framework import AgentFunctionApp
+
+
+def create_chat_client():
+    from agent_framework.openai import OpenAIChatClient
+
+    return OpenAIChatClient()
+
+
+app = AgentFunctionApp(client_factory=create_chat_client)
+
+
+@app.route(route="orders", methods=["POST"])
+@app.markdown_agent(arg_name="agent", agent_name="orders")
+async def process_order(req: func.HttpRequest, agent: Agent):
+    response = await agent.run(req.get_body().decode())
+    return response.text
+```
+
+`AgentFunctionApp` is owned by this extension and subclasses
+`azure.functions.FunctionApp`. The Azure Functions SDK does not need Agent APIs
+or modifications. One app uses the Microsoft Agent Framework provider selected
+by this package.
+
+Place the complete instructions at `orders.agent.md` or
+`agents/orders.agent.md`. The file is raw UTF-8 text; no front matter or runtime
+configuration is interpreted.
+
+## Skills and MCP servers
+
+Skills and MCP servers are discovered automatically from the app root and are
+available to each Agent binding by default:
+
+```text
+skills/inventory/SKILL.md
+mcp.json
+```
+
+`SKILL.md` uses Agent Skills frontmatter:
+
+```markdown
+---
+name: inventory
+description: Look up inventory policy and warehouse constraints.
+---
+
+Use the references in this skill when assessing stock.
+```
+
+The base extension discovers Skill directory paths without reading their
+contents. Microsoft Agent Framework parses and validates each `SKILL.md` when
+it loads the file-based Skills provider.
+
+V1 MCP discovery supports remote HTTP transports only:
+
+```json
+{
+    "servers": {
+        "inventory": {
+            "type": "streamable-http",
+            "url": "$INVENTORY_MCP_URL",
+            "tools": ["lookup_stock", "reserve_stock"],
+            "headers": {"X-Tenant": "%TENANT_ID%"},
+            "auth": {
+                "scope": "$INVENTORY_MCP_SCOPE",
+                "client_id": "%AZURE_CLIENT_ID%"
+            }
+        }
+    }
+}
+```
+
+`$VAR` and `%VAR%` references are resolved for each invocation, not during
+discovery. Missing values fail before connecting. Servers configured with
+headers or Entra authentication must use HTTPS; HTTP is accepted only for
+loopback development. Exposed MCP tool names are prefixed with the server name
+to prevent collisions between servers. Credentials, tokens, HTTP clients, MCP
+tools, and Agents are fresh invocation-owned resources and are closed on
+success, error, or cancellation. Do not place secrets directly in
+source-controlled `mcp.json`; use environment references.
+
+Every Agent in the Function App receives all valid Skills and MCP servers
+discovered from the app root:
+
+```python
+from azurefunctions.agents.extensions.agent_framework import AgentFunctionApp
+
+app = AgentFunctionApp(client_factory=create_chat_client)
+
+
+@app.markdown_agent(arg_name="agent", agent_name="orders")
+async def process_order(agent: Agent):
+        ...
+```
+
+V1 has no app-level or per-binding capability selectors. Skill scripts and MCP
+tools can perform privileged operations, so placing a definition under the app
+root grants every Agent in that app access to it. Use separate Function Apps
+when capabilities require isolation. Python `tools=` remain explicit because
+they are supplied directly to the Microsoft Agent Framework Agent.
+
+The constructor and decorator expose only `client_factory` and explicit Python
+`tools` in V1. The extension owns the Agent client, name, instructions, and
+discovered Skills/MCP integration. Configure `app_root` only when constructing
+`AgentFunctionApp`; decorators do not override it.
+
+## Durable Agents
+
+Durable orchestration support is optional:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[durable]"
+```
+
+Use `AgentFunctionApp` and call `context.call_agent(agent_name, input_)` inside a
+synchronous generator orchestrator. Agent execution is isolated in an activity
+so replay performs no nondeterministic work. Importing the package remains safe
+without Durable installed; using a Durable decorator requires the `[durable]`
+extra.
+
+All `call_agent()` invocations use the provider configured by `AgentFunctionApp`.
+They also use the app-level `skills` and `mcp_servers` defaults. V1 does not
+support selecting another provider or capability set from an orchestrator, and
+the schema-v1 orchestration payload contains no capability paths, settings, or
+secrets.

azurefunctions_agents_extensions_agent_framework-1.0.0b1/README.md

@@ -0,0 +1,157 @@
+# Azure Functions Microsoft Agent Framework Extension
+
+Inject Microsoft Agent Framework Agents built from raw `.agent.md` instructions
+into Python Azure Functions.
+
+## Install
+
+```text
+pip install azurefunctions-agents-extensions-agent-framework
+```
+
+Install Durable Functions support with the durable extra:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[durable]"
+```
+
+
+Install remote MCP transport and Entra support with the MCP extra:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[mcp]"
+```
+
+## Use an Agent app
+
+Create a zero-argument factory that returns a fresh MAF chat client. A new
+client and Agent context are created and closed for every Function invocation.
+
+```python
+import azure.functions as func
+from agent_framework import Agent
+from azurefunctions.agents.extensions.agent_framework import AgentFunctionApp
+
+
+def create_chat_client():
+    from agent_framework.openai import OpenAIChatClient
+
+    return OpenAIChatClient()
+
+
+app = AgentFunctionApp(client_factory=create_chat_client)
+
+
+@app.route(route="orders", methods=["POST"])
+@app.markdown_agent(arg_name="agent", agent_name="orders")
+async def process_order(req: func.HttpRequest, agent: Agent):
+    response = await agent.run(req.get_body().decode())
+    return response.text
+```
+
+`AgentFunctionApp` is owned by this extension and subclasses
+`azure.functions.FunctionApp`. The Azure Functions SDK does not need Agent APIs
+or modifications. One app uses the Microsoft Agent Framework provider selected
+by this package.
+
+Place the complete instructions at `orders.agent.md` or
+`agents/orders.agent.md`. The file is raw UTF-8 text; no front matter or runtime
+configuration is interpreted.
+
+## Skills and MCP servers
+
+Skills and MCP servers are discovered automatically from the app root and are
+available to each Agent binding by default:
+
+```text
+skills/inventory/SKILL.md
+mcp.json
+```
+
+`SKILL.md` uses Agent Skills frontmatter:
+
+```markdown
+---
+name: inventory
+description: Look up inventory policy and warehouse constraints.
+---
+
+Use the references in this skill when assessing stock.
+```
+
+The base extension discovers Skill directory paths without reading their
+contents. Microsoft Agent Framework parses and validates each `SKILL.md` when
+it loads the file-based Skills provider.
+
+V1 MCP discovery supports remote HTTP transports only:
+
+```json
+{
+    "servers": {
+        "inventory": {
+            "type": "streamable-http",
+            "url": "$INVENTORY_MCP_URL",
+            "tools": ["lookup_stock", "reserve_stock"],
+            "headers": {"X-Tenant": "%TENANT_ID%"},
+            "auth": {
+                "scope": "$INVENTORY_MCP_SCOPE",
+                "client_id": "%AZURE_CLIENT_ID%"
+            }
+        }
+    }
+}
+```
+
+`$VAR` and `%VAR%` references are resolved for each invocation, not during
+discovery. Missing values fail before connecting. Servers configured with
+headers or Entra authentication must use HTTPS; HTTP is accepted only for
+loopback development. Exposed MCP tool names are prefixed with the server name
+to prevent collisions between servers. Credentials, tokens, HTTP clients, MCP
+tools, and Agents are fresh invocation-owned resources and are closed on
+success, error, or cancellation. Do not place secrets directly in
+source-controlled `mcp.json`; use environment references.
+
+Every Agent in the Function App receives all valid Skills and MCP servers
+discovered from the app root:
+
+```python
+from azurefunctions.agents.extensions.agent_framework import AgentFunctionApp
+
+app = AgentFunctionApp(client_factory=create_chat_client)
+
+
+@app.markdown_agent(arg_name="agent", agent_name="orders")
+async def process_order(agent: Agent):
+        ...
+```
+
+V1 has no app-level or per-binding capability selectors. Skill scripts and MCP
+tools can perform privileged operations, so placing a definition under the app
+root grants every Agent in that app access to it. Use separate Function Apps
+when capabilities require isolation. Python `tools=` remain explicit because
+they are supplied directly to the Microsoft Agent Framework Agent.
+
+The constructor and decorator expose only `client_factory` and explicit Python
+`tools` in V1. The extension owns the Agent client, name, instructions, and
+discovered Skills/MCP integration. Configure `app_root` only when constructing
+`AgentFunctionApp`; decorators do not override it.
+
+## Durable Agents
+
+Durable orchestration support is optional:
+
+```text
+pip install "azurefunctions-agents-extensions-agent-framework[durable]"
+```
+
+Use `AgentFunctionApp` and call `context.call_agent(agent_name, input_)` inside a
+synchronous generator orchestrator. Agent execution is isolated in an activity
+so replay performs no nondeterministic work. Importing the package remains safe
+without Durable installed; using a Durable decorator requires the `[durable]`
+extra.
+
+All `call_agent()` invocations use the provider configured by `AgentFunctionApp`.
+They also use the app-level `skills` and `mcp_servers` defaults. V1 does not
+support selecting another provider or capability set from an orchestrator, and
+the schema-v1 orchestration payload contains no capability paths, settings, or
+secrets.

azurefunctions_agents_extensions_agent_framework-1.0.0b1/azurefunctions/__init__.py

@@ -0,0 +1 @@
+__path__ = __import__("pkgutil").extend_path(__path__, __name__)

azurefunctions_agents_extensions_agent_framework-1.0.0b1/azurefunctions/agents/__init__.py

@@ -0,0 +1 @@
+__path__ = __import__("pkgutil").extend_path(__path__, __name__)

azurefunctions_agents_extensions_agent_framework-1.0.0b1/azurefunctions/agents/extensions/__init__.py

@@ -0,0 +1 @@
+__path__ = __import__("pkgutil").extend_path(__path__, __name__)

azurefunctions_agents_extensions_agent_framework-1.0.0b1/azurefunctions/agents/extensions/agent_framework/__init__.py

@@ -0,0 +1,13 @@
+from azurefunctions.agents.extensions.base.durable import DurableAgentContext
+
+from .apps import AgentFunctionApp
+from .provider import AGENT_FRAMEWORK_PROVIDER_ID, ClientFactory
+
+__all__ = [
+    "AGENT_FRAMEWORK_PROVIDER_ID",
+    "AgentFunctionApp",
+    "ClientFactory",
+    "DurableAgentContext",
+]
+
+__version__ = '1.0.0b1'

azurefunctions_agents_extensions_agent_framework-1.0.0b1/azurefunctions/agents/extensions/agent_framework/apps.py

@@ -0,0 +1,103 @@
+from __future__ import annotations
+
+import os
+from collections.abc import Callable, Sequence
+from typing import Any, TypeVar
+
+import azure.functions as func
+from agent_framework import ToolTypes
+
+from azurefunctions.agents.extensions.base import (
+    configure_app,
+    durable_orchestration_trigger,
+)
+from azurefunctions.agents.extensions.base import markdown_agent as base_markdown_agent
+
+from .provider import AGENT_FRAMEWORK_PROVIDER_ID, ClientFactory
+
+_F = TypeVar("_F", bound=Callable[..., Any])
+
+
+def _provider_options(
+    *,
+    client_factory: ClientFactory | None = None,
+    tools: (
+        ToolTypes | Callable[..., Any] | Sequence[ToolTypes | Callable[..., Any]] | None
+    ) = None,
+) -> dict[str, object]:
+    options: dict[str, object] = {}
+    if client_factory is not None:
+        options["client_factory"] = client_factory
+    if tools is not None:
+        options["tools"] = tools
+    return options
+
+
+class _AgentFrameworkAppMixin:
+    def markdown_agent(
+        self,
+        *,
+        arg_name: str,
+        agent_name: str,
+        client_factory: ClientFactory | None = None,
+        tools: (
+            ToolTypes
+            | Callable[..., Any]
+            | Sequence[ToolTypes | Callable[..., Any]]
+            | None
+        ) = None,
+    ) -> Callable[[_F], _F]:
+        return base_markdown_agent(
+            self,
+            provider=AGENT_FRAMEWORK_PROVIDER_ID,
+            arg_name=arg_name,
+            agent_name=agent_name,
+            **_provider_options(client_factory=client_factory, tools=tools),
+        )
+
+
+class AgentFunctionApp(
+    _AgentFrameworkAppMixin,
+    func.FunctionApp,
+):
+    """Azure Functions app configured for Microsoft Agent Framework Agents."""
+
+    def __init__(
+        self,
+        *,
+        client_factory: ClientFactory,
+        app_root: str | os.PathLike[str] | None = None,
+        tools: (
+            ToolTypes
+            | Callable[..., Any]
+            | Sequence[ToolTypes | Callable[..., Any]]
+            | None
+        ) = None,
+        http_auth_level: func.AuthLevel | str = func.AuthLevel.FUNCTION,
+    ) -> None:
+        super().__init__(
+            http_auth_level=http_auth_level,
+        )
+        configure_app(
+            self,
+            provider=AGENT_FRAMEWORK_PROVIDER_ID,
+            app_root=app_root,
+            provider_options=_provider_options(
+                client_factory=client_factory,
+                tools=tools,
+            ),
+        )
+
+    def orchestration_trigger(
+        self,
+        context_name: str,
+        orchestration: str | None = None,
+        input_type: type | None = None,
+    ) -> Callable[..., Any]:
+        return durable_orchestration_trigger(
+            self,
+            sdk_decorator=super().orchestration_trigger,
+            context_name=context_name,
+            orchestration=orchestration,
+            input_type=input_type,
+        )