feat: Implement Gradio application with interactive query and expiry sweep, and configure MCP server support.

README.md

@@ -1,6 +1,109 @@
-# CredentialWatch Agent
+---
+title: CredentialWatch MCP Server
+emoji: 🩺
+colorFrom: blue
+colorTo: purple
+sdk: gradio
+python_version: 3.10
+sdk_version: 6.0.0
+app_file: src/credentialwatch_agent/main.py
+short_description: "MCP-enabled Gradio Space for healthcare credential monitoring."
+tags:
+  - mcp
+  - gradio
+  - healthcare
+  - tools
+pinned: false
+---
+
+# CredentialWatch MCP Server
+
+Agent-ready Gradio Space that exposes healthcare credential tools (lookups, expiry checks, risk scoring) over **Model Context Protocol (MCP)**.
+
+## Hugging Face Space
+
+- **SDK**: Gradio
+- **Entry file**: `src/credentialwatch_agent/main.py`
+- **Python**: 3.10+
+
+Deploy this repo as a Gradio Space and it will automatically serve both the web UI and an MCP server.
+
+## MCP Server
+
+This Space exposes its tools via Model Context Protocol (MCP) using Gradio.
+
+### How MCP is enabled
+In `src/credentialwatch_agent/main.py` we:
+1. Install Gradio with MCP support: `pip install "gradio[mcp]"`
+2. Define typed Python functions with docstrings
+3. Launch the app with MCP support:
+   ```python
+   demo.launch(mcp_server=True)
+   ```
+
+Alternatively, you can set the environment variable:
+```bash
+export GRADIO_MCP_SERVER=True
+```
+
+### MCP endpoints
+When the Space is running, Gradio exposes:
+- **MCP SSE endpoint**: `https://<space-host>/gradio_api/mcp/sse`
+- **MCP schema**: `https://<space-host>/gradio_api/mcp/schema`
+
+## Using this Space from an MCP client
+
+### Easiest: Hugging Face MCP Server (no manual config)
+1. Go to your HF **MCP settings**: https://huggingface.co/settings/mcp
+2. Add this Space under **Spaces Tools** (look for the MCP badge on the Space).
+3. Restart your MCP client (VS Code, Cursor, Claude Code, etc.).
+4. The tools from this Space will appear as MCP tools and can be called directly.
+
+### Manual config (generic MCP client using mcp-remote)
+If your MCP client uses a JSON config, you can point it to the SSE endpoint via `mcp-remote`:
+
+```jsonc
+{
+  "mcpServers": {
+    "credentialwatch": {
+      "command": "npx",
+      "args": [
+        "mcp-remote",
+        "https://<space-host>/gradio_api/mcp/sse"
+      ]
+    }
+  }
+}
+```
+Replace `<space-host>` with the full URL of your Space.
+
+## Local development
+
+```bash
+# 1. Install deps
+uv sync
+
+# 2. Run locally
+uv run python src/credentialwatch_agent/main.py
+# or
+GRADIO_MCP_SERVER=True uv run python src/credentialwatch_agent/main.py
+```
+
+The local server will be available at `http://127.0.0.1:7860`, and MCP at `http://127.0.0.1:7860/gradio_api/mcp/sse`.
+
+## Deploying to Hugging Face Spaces
 
-MCP-powered agent system for monitoring healthcare provider credentials. This system uses LangGraph to orchestrate workflows for checking credential expiry and answering interactive queries, leveraging Model Context Protocol (MCP) to connect to external data sources.
+1. Create a new Space with SDK = **Gradio**.
+2. Push this repo to the Space (Git or `huggingface_hub`).
+3. Ensure the YAML header in `README.md` is present and correct.
+4. Wait for the Space to build and start — it should show an **MCP badge** automatically.
+
+## Troubleshooting
+
+- **README.md location**: Must be at the repo root and named `README.md` (all caps).
+- **YAML Header**: Must be the very first thing in the file, delimited by `---`.
+- **Configuration Error**: Check `sdk`, `app_file`, and `python_version` in the YAML.
+- **MCP Badge Missing**: Ensure `demo.launch(mcp_server=True)` is called or `GRADIO_MCP_SERVER=True` is set, and the Space is public.
 
 ## Features
 
@@ -39,20 +142,6 @@ CRED_DB_MCP_URL=http://localhost:8002/sse
 ALERT_MCP_URL=http://localhost:8003/sse
 ```
 
-## Running the Application
-
-To run the agent system with the Gradio UI using `uv`:
-
-```bash
-uv run python src/credentialwatch_agent/main.py
-```
-OR
-```bash
-uv run -m credentialwatch_agent.main
-```
-
-The UI will be available at `http://localhost:7860`.
-
 ## Architecture
 
 - **`src/credentialwatch_agent/agents/`**: Contains LangGraph workflow definitions.

pyproject.toml

@@ -9,7 +9,7 @@ dependencies = [
     "langchain-openai>=0.0.5",
     "pydantic>=2.0.0",
     "mcp>=0.1.0",
-    "gradio>=4.0.0",
+    "gradio[mcp]>=4.0.0",
     "python-dotenv>=1.0.0",
     "httpx>=0.25.0" 
 ]

src/credentialwatch_agent/main.py

@@ -105,7 +105,7 @@ if __name__ == "__main__":
     
     try:
         loop.run_until_complete(mcp_client.connect())
-        demo.launch(server_name="0.0.0.0", server_port=7860)
+        demo.launch(server_name="0.0.0.0", server_port=7860, mcp_server=True)
     except KeyboardInterrupt:
         pass
     finally:

uv.lock

@@ -338,7 +338,7 @@ name = "credentialwatch-agent"
 version = "0.1.0"
 source = { editable = "." }
 dependencies = [
-    { name = "gradio" },
+    { name = "gradio", extra = ["mcp"] },
     { name = "httpx" },
     { name = "langchain-openai" },
     { name = "langgraph" },
@@ -349,7 +349,7 @@ dependencies = [
 
 [package.metadata]
 requires-dist = [
-    { name = "gradio", specifier = ">=4.0.0" },
+    { name = "gradio", extras = ["mcp"], specifier = ">=4.0.0" },
     { name = "httpx", specifier = ">=0.25.0" },
     { name = "langchain-openai", specifier = ">=0.0.5" },
     { name = "langgraph", specifier = ">=0.0.10" },
@@ -510,6 +510,12 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/09/21/27ae5f4b2191a5d58707fc610e67453781a2b948a675a7cf06c99497ffa1/gradio-6.0.1-py3-none-any.whl", hash = "sha256:0f98dc8b414a3f3773cbf3caf5a354507c8ae309ed8266e2f30ca9fa53f379b8", size = 21559963, upload-time = "2025-11-25T18:44:01.544Z" },
 ]
 
+[package.optional-dependencies]
+mcp = [
+    { name = "mcp" },
+    { name = "pydantic" },
+]
+
 [[package]]
 name = "gradio-client"
 version = "2.0.0"