Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@northstar-clinic Show available appointment slots"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
This branch extends the existing appointment application with text-first MCP tools and an interactive MCP App that compatible hosts such as Claude Desktop can render inside the conversation.
The teaching demo uses an in-memory AppointmentStore so it can be reset
instantly. In a production application, this module would be replaced by a
database-backed repository without changing the browser-to-API contract.
Start the clinic API and keep it running:
uv sync
uv run uvicorn clinic_api:app --host 127.0.0.1 --port 8765
uv run uvicorn clinic_api:app --host 127.0.0.1 --port 8765The browser version is at http://127.0.0.1:8765. API docs are at http://127.0.0.1:8765/docs.
Related MCP server: Ecuro Light API MCP Server
clinic_browser.html is the browser UI.
clinic_api.py exposes appointment and booking HTTP endpoints.
clinic_store.py owns availability and atomic booking state.
clinic_mcp.py adapts those HTTP capabilities into MCP tools and exposes the
widget resource.
clinic_widget.html is the interactive UI rendered by the MCP host. It calls
MCP tools through the MCP Apps SDK; it does not bypass the MCP server to call
FastAPI directly.
Claude Desktop starts clinic_mcp.py as a local stdio MCP server. After changing
the MCP server or widget, fully quit and reopen Claude Desktop so it reconnects
and reloads the resource.
Open the Claude Desktop app.
Press Command + , to open Settings.
In the settings sidebar, find Desktop app and select Developer.
Under Local MCP servers, click Edit Config. Claude opens its
claude_desktop_config.json file in the default editor.
Add the northstar-clinic entry under mcpServers:
{
"mcpServers": {
"northstar-clinic": {
"command": "/Users/yashpatil/.local/bin/uv",
"args": [
"run",
"--directory",
"/Users/yashpatil/Developer/AI/mcp_apps",
"python",
"clinic_mcp.py"
],
"env": {
"CLINIC_API_URL": "http://127.0.0.1:8765"
}
}
}
}If the file already contains other MCP servers, add northstar-clinic inside
the existing mcpServers object instead of replacing the entire file.
Save the file, fully quit Claude Desktop with Command + Q, and reopen it.
Return to Settings → Desktop app → Developer and confirm that
northstar-clinic appears under Local MCP servers with a running status.
The FastAPI backend must be running separately before using the MCP tools:
uv run uvicorn clinic_api:app --host 127.0.0.1 --port 8765Ask: Use Northstar Clinic to find all available appointments and show the interactive dashboard.
Filter to Dermatology in the rendered widget.
Click Book visit and review the appointment.
Enter the required contact number and reason, then click Confirm appointment.
Show that the appointment disappears from availability and that the receipt contains its reference.
Verify the same change with GET /api/appointments or the browser frontend.
git switch main # existing FastAPI + browser application
git switch codex/mcp-app # MCP server + Claude-rendered widget
git diff main..codex/mcp-appcurl http://127.0.0.1:8765/health
curl http://127.0.0.1:8765/api/appointmentsThe API is intentionally in-memory. Restarting it restores the six seeded slots.
Book one available appointment for a patient.
| Name | Required | Description | Default |
|---|---|---|---|
| patient_name | Yes | ||
| visit_reason | Yes | ||
| contact_phone | Yes | ||
| appointment_id | Yes | ||
| additional_notes | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations exist, so the description must disclose behavior. It states 'Book one available appointment' implying availability checking, but doesn't explain side effects, success/failure behavior, or permission requirements. This leaves significant gaps.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
One sentence, no redundant words. However, it may be too terse to provide needed context, but as a concise statement it's well-structured.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With no annotations, no output schema, and 5 params, the single-sentence description is insufficient. It lacks parameter details, usage scenarios, and behavioral outcomes, making it incomplete for an agent to invoke correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema has 0% description coverage, so the description must compensate. It implies appointment_id is the slot identifier and patient fields are the booking details, but doesn't clarify formats, defaults, or additional_notes. This is minimal compensation.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses the specific verb 'book' with resource 'appointment' and recipient 'patient', clearly conveying the action. It distinguishes from siblings like find_appointments (searching) and reset_demo_data (resetting).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit when-to-use or alternative guidance is provided. The purpose implies use when booking an appointment, but no exclusions or comparisons are mentioned. Sibling tools are not referenced.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Find available Northstar Clinic appointments and show an interactive booking dashboard.
| Name | Required | Description | Default |
|---|---|---|---|
| date | No | ||
| specialty | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description must carry the full burden of behavioral disclosure. It does reveal a notable behavior—showing an interactive booking dashboard—which goes beyond a plain data retrieval. However, it does not mention permissions, side effects (e.g., does not book), or any other behavioral constraints, leaving gaps.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence with no redundant wording, effectively stating both purpose and a key behavior. It is appropriately concise, though given the sparse schema and annotations, it could have used a bit more detail without sacrificing conciseness.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
The context is incomplete: there is no output schema, no annotations, and no parameter descriptions. The description only covers the tool's purpose and the dashboard behavior, leaving the agent uncertain about filter semantics, return format, and potential side effects. For a simple two-parameter tool, this is still insufficient.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The schema has no descriptions for the two parameters, and schema description coverage is 0%. The description does not explain the 'date' and 'specialty' parameters, their accepted formats, or how they filter results. Only the parameter names give a hint, but no additional meaning is added.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('find') and resource ('available Northstar Clinic appointments'), and adds a distinctive result ('show an interactive booking dashboard'). This clearly distinguishes it from sibling tools like book_appointment, which handles booking, and reset_demo_data, which resets demo data.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies its usage is for finding appointments, but it does not explicitly state when to use this tool versus book_appointment or reset_demo_data. There are no exclusions, prerequisites, or alternative mentions, making the guidance only implicit rather than explicit.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Restore all seeded appointments so the demo can be repeated.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It discloses that the tool restores seeded appointments, implying a write/overwrite operation, but it does not explicitly mention that current changes will be lost or that the operation is irreversible. This is a moderate level of transparency for a reset tool.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence, front-loaded with the action and purpose, with no redundant information. It earns its place perfectly.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple, parameterless tool, the description covers the core action and the intended use case. However, it omits potential side effects (e.g., overwriting all existing appointments) and return value details, which would make it fully complete. Given the simplicity, the gap is small.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The tool has zero parameters, and the description does not need to explain parameter meanings. Per the rubric, the baseline for 0 parameters is 4, and the description adequately covers the tool's purpose without parameter ambiguity.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the action ('Restore all seeded appointments') and the resource ('seeded appointments'), with the purpose of repeating the demo. This distinguishes it from sibling tools like find_appointments and book_appointment, which serve different functions.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The phrase 'so the demo can be repeated' clearly indicates when this tool should be used: to reset demo data after a demo session. Although no explicit exclusions or alternative tool references are given, the context is clear and it implicitly contrasts with find/book operations.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Each tool has a distinct purpose: find appointments, book an appointment, and reset demo data. There is no overlap or ambiguity between them.
All tools follow a consistent verb_noun pattern with lowercase and underscores. The singular/plural differences are natural and do not break the convention.
Three tools is a well-scoped count for a clinic demo server, covering the essential booking flow without unnecessary complexity.
The toolkit covers finding and booking appointments but lacks cancellation, rescheduling, or listing booked appointments, which are typical in appointment management. The demo reset helps but does not fully fill these gaps.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Read appointments, types, calendars and availability; create, cancel or reschedule bookings.
AI-native scheduling: check availability, book meetings, cancel and reschedule via MCP
Hosted MCP server for Cliniko — patients, appointments, availability, and invoices for AI agents.
Scheduling, availability, clients, billing and CRM for appointment-based services.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/yashprogrammer/mcp-apps'
If you have feedback or need assistance with the MCP directory API, please join our Discord server