uoft-mcp 0.1.0__tar.gz

uoft_mcp-0.1.0/.github/workflows/publish.yml

@@ -0,0 +1,69 @@
+name: Publish to PyPI
+
+on:
+  push:
+    tags:
+      - "v*"
+
+jobs:
+  build:
+    name: Test and build distributions
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    steps:
+      - name: Check out repository
+        uses: actions/checkout@v7
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
+        with:
+          enable-cache: true
+
+      - name: Install Python
+        run: uv python install 3.13
+
+      - name: Install dependencies
+        run: uv sync --locked --managed-python
+
+      - name: Run tests
+        run: uv run --locked pytest -q
+
+      - name: Run lint and format checks
+        run: |
+          uv run --locked ruff check .
+          uv run --locked ruff format --check .
+
+      - name: Build distributions
+        run: uv build
+
+      - name: Upload distributions
+        uses: actions/upload-artifact@v7
+        with:
+          name: dist
+          path: dist/
+          if-no-files-found: error
+
+  publish:
+    name: Publish distributions
+    needs: build
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/project/uoft-mcp/
+    permissions:
+      id-token: write
+    steps:
+      - name: Install uv
+        uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
+        with:
+          enable-cache: false
+
+      - name: Download distributions
+        uses: actions/download-artifact@v8
+        with:
+          name: dist
+          path: dist/
+
+      - name: Publish to PyPI
+        run: uv publish --trusted-publishing always

uoft_mcp-0.1.0/.gitignore

@@ -0,0 +1,26 @@
+# Python environments and local dependency caches
+.venv/
+.venv-*/
+.uv-cache/
+.python/
+__pycache__/
+*.py[cod]
+
+# Test, lint, and build output
+.pytest_cache/
+.ruff_cache/
+.coverage
+htmlcov/
+build/
+dist/
+*.egg-info/
+
+# Local configuration and editor state
+.env
+.env.*
+!.env.example
+.vscode/
+.idea/
+*.log
+.DS_Store
+Thumbs.db

uoft_mcp-0.1.0/.python-version

@@ -0,0 +1 @@
+3.13

uoft_mcp-0.1.0/EXPLAINED.md

@@ -0,0 +1,111 @@
+# How This MCP Wrapper Works
+
+## The Basic Idea
+
+An MCP client, such as an assistant application, launches this Python process and
+asks it to call named tools. Our tools make ordinary HTTP requests to UofT and send
+the JSON response back to the client. The MCP SDK handles the protocol, tool
+discovery, and argument validation.
+
+For example:
+
+```text
+MCP client calls get_course_details(course_code="CSC108H1", section_code="F")
+  -> server.py maps the tool arguments to an HTTP request
+  -> client.py sends GET /ttb/getCoursesByCodeAndSectionCode/CSC108H1?sectionCode=F
+  -> UofT returns JSON containing course and section information
+  -> server.py returns that JSON as one MCP text block
+```
+
+There is no local course catalog, AI model, or scheduling algorithm in this project.
+The timetable information comes directly from UofT on each call.
+
+## Files You Will Work With
+
+| File | Responsibility |
+| --- | --- |
+| `uoft_mcp/server.py` | Defines the seven tools, maps arguments, and runs the MCP server. |
+| `uoft_mcp/client.py` | Sets the API URL, request headers, timeout, and HTTP error handling. |
+| `uoft_mcp/__main__.py` | Makes `python -m uoft_mcp` call the server's `main()` function. |
+| `uoft_mcp/__init__.py` | Marks the directory as an importable package. |
+| `pyproject.toml` | Package metadata, dependencies, command entry point, and test/lint settings. |
+| `uv.lock` | Exact resolved dependencies for reproducible installation; generated by uv. |
+| `.python-version` | Tells uv to use Python 3.13 for this project. |
+| `.gitignore` | Keeps environments, caches, local secrets, and generated files out of Git. |
+| `tests/` | Offline tests of the API mappings and MCP interface, including real process pipes. |
+| `timetable_builder.json` | Your original endpoint reference; not loaded at runtime. |
+
+Runtime dependencies are `mcp` (the protocol), `httpx` (HTTP), and `pydantic`
+(argument constraints). `pytest` and `ruff` are development tools. Other packages in
+the lockfile are dependencies of those libraries.
+
+## Reading a Tool
+
+Look at `search_departments` in `server.py`. Its decorator registers the function as
+an MCP tool. Its docstring describes the tool to clients, and its type annotations
+become the tool's input schema.
+
+The `ctx` parameter is injected by the SDK, so users do not supply it. It gives the
+function access to the shared HTTP client. The remaining arguments are the public
+inputs: `term` and `divisions`.
+
+The function calls `_request` with the endpoint and query parameters. More involved
+tools follow the same pattern: `search_course_titles` changes `lower_threshold` into
+the API's `lowerThreshold`, while `search_courses` builds a POST body with nested
+`courseCodeAndTitleProps` and the documented filters.
+
+`search_courses` uses the live frontend's page-1 and ascending-sort defaults. It also
+includes `departmentProps: []`: omitting that field produced an HTTP 500 during
+verification. The original JSON is kept unchanged so this correction is explicit.
+
+## Startup, Responses, and Errors
+
+`create_server()` builds the server without making an API request. Its lifespan
+opens one `httpx.AsyncClient` when the server starts, shares its connection pool
+across tool calls, and closes it when the server stops.
+
+Every request has a 30-second HTTPX timeout for network operations. Headers include
+JSON acceptance, a browser-like user agent, and the Timetable Builder origin and
+referrer. This public API does not need a token.
+
+`request_json()` parses the response but does not convert it into a custom course
+class. `_request()` serializes that data into one JSON text block. This preserves
+top-level arrays and objects uniformly; programmatic MCP clients can use
+`json.loads(result.content[0].text)`. There is no separate structured-output schema,
+because the upstream response shapes are intentionally left open.
+
+An HTTP error, timeout, connection failure, or invalid JSON response raises
+`TimetableAPIError`. The MCP layer translates it to `ToolError`, which the client
+receives as a failed tool result. HTTP errors retain the status code, but arbitrary
+upstream error bodies are not exposed. A successful HTTP response's application-level
+`status` is passed through unchanged.
+
+Do not add `print()` debugging to the server: stdout is its protocol connection.
+Use Python's `logging` module, which is configured to write to stderr.
+
+## Adding Another Endpoint
+
+1. Verify the endpoint and required parameters against the reference and live API.
+2. Add a typed, documented function inside `create_server()` with a `@server.tool`
+   decorator. For another lookup, follow an existing tool's read-only annotation
+   and `structured_output=False` setting.
+3. Map its arguments to a fixed path and query parameters or JSON body, then call
+   `_request`. Do not accept arbitrary destination URLs.
+4. Add a mocked MCP call in `tests/test_tools.py` that checks the outgoing request
+   and returned data. Update the expected tool count/name set and the README.
+5. Run the README's test and lint commands, then commit the completed change.
+
+Tests inject `httpx.MockTransport` through `create_server(transport=...)`. Requests
+still go through the SDK, tool functions, and HTTP client, but the transport returns
+controlled responses without contacting UofT. The stdio tests separately verify that
+an installed package starts correctly and stdout contains protocol JSON only.
+
+Catalog dumps, timetable sharing, and schedule generation are intentionally deferred.
+Saving a timetable would need different tool annotations, and generation needs a
+verified request schema. They should be added as explicit features when needed.
+
+## Milestones
+
+The initial work is split into local commits: project setup, working tools and tests,
+and these explanatory documents. Use `git log --oneline` to see the milestones and
+`git show <commit>` to inspect one. Nothing is pushed to a remote repository.

uoft_mcp-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 SleepyPandas
+
+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.

uoft_mcp-0.1.0/PKG-INFO

@@ -0,0 +1,164 @@
+Metadata-Version: 2.5
+Name: uoft-mcp
+Version: 0.1.0
+Summary: A small MCP wrapper for the UofT Timetable Builder API.
+Project-URL: Homepage, https://github.com/SleepyPandas/UofT-MCP
+Project-URL: Issues, https://github.com/SleepyPandas/UofT-MCP/issues
+Project-URL: Repository, https://github.com/SleepyPandas/UofT-MCP
+Author: SleepyPandas
+License-Expression: MIT
+License-File: LICENSE
+Keywords: mcp,timetable,university-of-toronto,uoft
+Requires-Python: >=3.13
+Requires-Dist: httpx<1,>=0.28
+Requires-Dist: mcp<3,>=2
+Requires-Dist: pydantic<3,>=2.11
+Description-Content-Type: text/markdown
+
+# UofT Timetable Builder MCP
+
+A small Python MCP server for the public [UofT Timetable Builder](https://ttb.utoronto.ca/)
+API. It exposes seven course-lookup tools over local stdio using the
+[official MCP Python SDK](https://github.com/modelcontextprotocol/python-sdk).
+
+No API key, database, web server, or environment variables are required. This is an
+unofficial wrapper; it does not enroll students, build schedules, or save timetables.
+
+## Connect an MCP Client
+
+Install [uv](https://docs.astral.sh/uv/getting-started/installation/), then add this
+configuration to any client that supports `mcpServers`:
+
+```json
+{
+  "mcpServers": {
+    "uoft-timetable": {
+      "command": "uvx",
+      "args": ["uoft-mcp@latest"],
+      "env": {
+        "UV_HTTP_TIMEOUT": "300"
+      }
+    }
+  }
+}
+```
+
+Restart the client after saving its configuration. On Windows, if the client cannot
+find `uvx`, restart it after installing uv or replace `"uvx"` with the absolute path
+reported by `where.exe uvx`.
+
+`uvx` downloads the published package into an isolated environment and starts the
+`uoft-mcp` command. No repository clone, virtual environment setup, API key, server
+URL, or listening port is needed. The first start can take longer while uv downloads
+Python and the dependencies; later starts use its cache.
+
+To pin a release instead of following the newest release, use
+`"args": ["uoft-mcp==0.1.0"]`.
+
+## Local Development
+
+From a clone of this repository:
+
+```powershell
+uv python install 3.13
+uv sync --locked --managed-python
+```
+
+`uv sync` creates `.venv`, installs the package and development tools, and uses the
+committed `uv.lock`. `.python-version` selects Python 3.13; the package supports
+Python 3.13 and newer.
+
+Run the local checkout with:
+
+```powershell
+uv run --locked python -m uoft_mcp
+```
+
+The installed `uoft-mcp` command is another entry point. The process waits for an
+MCP client on stdin; a blank terminal is expected. Use Ctrl+C to stop a manual run.
+Stdout carries protocol messages only, and logging goes to stderr.
+
+## Tools
+
+| Tool | Arguments and purpose |
+| --- | --- |
+| `get_current_sessions` | No arguments. Get current session IDs; skip entries with `header: true`. |
+| `get_reference_data` | No arguments. Get campus, division, delivery-mode, and sorting values. |
+| `get_divisions` | No arguments. List recognized faculty/division codes. |
+| `search_departments` | Required `term` keyword and `divisions` code. |
+| `search_course_titles` | Required `term`, `divisions`, and `sessions` strings. Optional `lower_threshold=50`, `upper_threshold=200`. |
+| `get_course_details` | Required `course_code`; optional `section_code` of `F`, `S`, or `Y`. |
+| `search_courses` | Optional code/title, section, description, division, session, campus, delivery, and pagination filters. |
+
+Each successful tool returns one text block containing the complete upstream JSON.
+The wrapper preserves fields and arrays, including upstream `payload` and `status`
+envelopes. It does not summarize, truncate, or reshape course data.
+
+### Example Workflow
+
+1. Call `get_current_sessions` with `{}` and select a non-header entry's `value`.
+2. Call `get_divisions` or `get_reference_data` for valid filter codes.
+3. Call `search_course_titles` with these arguments, substituting the session value:
+
+```json
+{"term": "CSC108", "divisions": "ARTSC", "sessions": "SESSION_ID_FROM_STEP_1"}
+```
+
+4. Use the returned exact course code in `get_course_details`:
+
+```json
+{"course_code": "CSC108H1", "section_code": "F"}
+```
+
+5. For filtered, paginated results, call `search_courses`:
+
+```json
+{
+  "course_code": "CSC108H1",
+  "divisions": ["ARTSC"],
+  "sessions": ["SESSION_ID_FROM_STEP_1"],
+  "page": 1,
+  "page_size": 2
+}
+```
+
+`search_courses` also accepts `course_title`, `course_section_code`,
+`search_course_description`, `campuses`, `delivery_modes`, and `direction` (`asc` or
+`desc`). Pages start at **1**, page size defaults to **20**, and sorting defaults to
+`asc`. Omitted collection filters become empty arrays. Course codes should be exact;
+use autocomplete for prefixes or `course_title` for keyword searches.
+
+## Checks
+
+```powershell
+uv run --locked pytest -q
+uv run --locked ruff check .
+uv run --locked ruff format --check .
+```
+
+The tests run offline. They cover all seven tools, request mapping, raw JSON
+preservation, validation, HTTP errors, timeouts, connection errors, invalid JSON,
+shared-client cleanup, MCP discovery, and actual stdio subprocesses.
+
+Initial verification: 33 tests passed, and all seven tools returned successful live
+responses from UofT, including an exact-code search with `page_size=2`. Live requests
+are deliberately not part of the test suite, so tests remain reproducible.
+
+## API Notes
+
+- The supplied [timetable_builder.json](timetable_builder.json) remains the original
+  reference. Live checks found two missing details: pagination starts at 1, and
+  paginated search requires an empty `departmentProps` array when not filtering by
+  department. The wrapper supplies it.
+- Get division codes from the API. For example, the live API uses `ERIN` and `SCAR`
+  for Mississauga and Scarborough, rather than the reference's `UTM` and `UTSC` examples.
+- Even one course can have a large response because all its sections are included.
+  Choose narrow filters and small page sizes. The wrapper never fetches extra pages.
+- HTTP failures become MCP tool errors containing the endpoint and status code.
+  UofT may return HTTP 404 for no matching courses. Timeouts, connection failures,
+  and malformed JSON get their own readable errors. No automatic retries occur.
+- This API is not covered by an official support guarantee. Changes upstream may
+  require updating the mappings. Successful HTTP responses are preserved as supplied,
+  including any application-level status messages inside their JSON.
+
+For a walkthrough of the code and how to extend it, read [EXPLAINED.md](EXPLAINED.md).

uoft_mcp-0.1.0/README.md

@@ -0,0 +1,147 @@
+# UofT Timetable Builder MCP
+
+A small Python MCP server for the public [UofT Timetable Builder](https://ttb.utoronto.ca/)
+API. It exposes seven course-lookup tools over local stdio using the
+[official MCP Python SDK](https://github.com/modelcontextprotocol/python-sdk).
+
+No API key, database, web server, or environment variables are required. This is an
+unofficial wrapper; it does not enroll students, build schedules, or save timetables.
+
+## Connect an MCP Client
+
+Install [uv](https://docs.astral.sh/uv/getting-started/installation/), then add this
+configuration to any client that supports `mcpServers`:
+
+```json
+{
+  "mcpServers": {
+    "uoft-timetable": {
+      "command": "uvx",
+      "args": ["uoft-mcp@latest"],
+      "env": {
+        "UV_HTTP_TIMEOUT": "300"
+      }
+    }
+  }
+}
+```
+
+Restart the client after saving its configuration. On Windows, if the client cannot
+find `uvx`, restart it after installing uv or replace `"uvx"` with the absolute path
+reported by `where.exe uvx`.
+
+`uvx` downloads the published package into an isolated environment and starts the
+`uoft-mcp` command. No repository clone, virtual environment setup, API key, server
+URL, or listening port is needed. The first start can take longer while uv downloads
+Python and the dependencies; later starts use its cache.
+
+To pin a release instead of following the newest release, use
+`"args": ["uoft-mcp==0.1.0"]`.
+
+## Local Development
+
+From a clone of this repository:
+
+```powershell
+uv python install 3.13
+uv sync --locked --managed-python
+```
+
+`uv sync` creates `.venv`, installs the package and development tools, and uses the
+committed `uv.lock`. `.python-version` selects Python 3.13; the package supports
+Python 3.13 and newer.
+
+Run the local checkout with:
+
+```powershell
+uv run --locked python -m uoft_mcp
+```
+
+The installed `uoft-mcp` command is another entry point. The process waits for an
+MCP client on stdin; a blank terminal is expected. Use Ctrl+C to stop a manual run.
+Stdout carries protocol messages only, and logging goes to stderr.
+
+## Tools
+
+| Tool | Arguments and purpose |
+| --- | --- |
+| `get_current_sessions` | No arguments. Get current session IDs; skip entries with `header: true`. |
+| `get_reference_data` | No arguments. Get campus, division, delivery-mode, and sorting values. |
+| `get_divisions` | No arguments. List recognized faculty/division codes. |
+| `search_departments` | Required `term` keyword and `divisions` code. |
+| `search_course_titles` | Required `term`, `divisions`, and `sessions` strings. Optional `lower_threshold=50`, `upper_threshold=200`. |
+| `get_course_details` | Required `course_code`; optional `section_code` of `F`, `S`, or `Y`. |
+| `search_courses` | Optional code/title, section, description, division, session, campus, delivery, and pagination filters. |
+
+Each successful tool returns one text block containing the complete upstream JSON.
+The wrapper preserves fields and arrays, including upstream `payload` and `status`
+envelopes. It does not summarize, truncate, or reshape course data.
+
+### Example Workflow
+
+1. Call `get_current_sessions` with `{}` and select a non-header entry's `value`.
+2. Call `get_divisions` or `get_reference_data` for valid filter codes.
+3. Call `search_course_titles` with these arguments, substituting the session value:
+
+```json
+{"term": "CSC108", "divisions": "ARTSC", "sessions": "SESSION_ID_FROM_STEP_1"}
+```
+
+4. Use the returned exact course code in `get_course_details`:
+
+```json
+{"course_code": "CSC108H1", "section_code": "F"}
+```
+
+5. For filtered, paginated results, call `search_courses`:
+
+```json
+{
+  "course_code": "CSC108H1",
+  "divisions": ["ARTSC"],
+  "sessions": ["SESSION_ID_FROM_STEP_1"],
+  "page": 1,
+  "page_size": 2
+}
+```
+
+`search_courses` also accepts `course_title`, `course_section_code`,
+`search_course_description`, `campuses`, `delivery_modes`, and `direction` (`asc` or
+`desc`). Pages start at **1**, page size defaults to **20**, and sorting defaults to
+`asc`. Omitted collection filters become empty arrays. Course codes should be exact;
+use autocomplete for prefixes or `course_title` for keyword searches.
+
+## Checks
+
+```powershell
+uv run --locked pytest -q
+uv run --locked ruff check .
+uv run --locked ruff format --check .
+```
+
+The tests run offline. They cover all seven tools, request mapping, raw JSON
+preservation, validation, HTTP errors, timeouts, connection errors, invalid JSON,
+shared-client cleanup, MCP discovery, and actual stdio subprocesses.
+
+Initial verification: 33 tests passed, and all seven tools returned successful live
+responses from UofT, including an exact-code search with `page_size=2`. Live requests
+are deliberately not part of the test suite, so tests remain reproducible.
+
+## API Notes
+
+- The supplied [timetable_builder.json](timetable_builder.json) remains the original
+  reference. Live checks found two missing details: pagination starts at 1, and
+  paginated search requires an empty `departmentProps` array when not filtering by
+  department. The wrapper supplies it.
+- Get division codes from the API. For example, the live API uses `ERIN` and `SCAR`
+  for Mississauga and Scarborough, rather than the reference's `UTM` and `UTSC` examples.
+- Even one course can have a large response because all its sections are included.
+  Choose narrow filters and small page sizes. The wrapper never fetches extra pages.
+- HTTP failures become MCP tool errors containing the endpoint and status code.
+  UofT may return HTTP 404 for no matching courses. Timeouts, connection failures,
+  and malformed JSON get their own readable errors. No automatic retries occur.
+- This API is not covered by an official support guarantee. Changes upstream may
+  require updating the mappings. Successful HTTP responses are preserved as supplied,
+  including any application-level status messages inside their JSON.
+
+For a walkthrough of the code and how to extend it, read [EXPLAINED.md](EXPLAINED.md).

uoft_mcp-0.1.0/pyproject.toml

@@ -0,0 +1,42 @@
+[build-system]
+requires = ["hatchling>=1.27,<2"]
+build-backend = "hatchling.build"
+
+[project]
+name = "uoft-mcp"
+version = "0.1.0"
+description = "A small MCP wrapper for the UofT Timetable Builder API."
+readme = "README.md"
+license = "MIT"
+authors = [{ name = "SleepyPandas" }]
+keywords = ["mcp", "uoft", "university-of-toronto", "timetable"]
+requires-python = ">=3.13"
+dependencies = [
+    "httpx>=0.28,<1",
+    "mcp>=2,<3",
+    "pydantic>=2.11,<3",
+]
+
+[project.urls]
+Homepage = "https://github.com/SleepyPandas/UofT-MCP"
+Issues = "https://github.com/SleepyPandas/UofT-MCP/issues"
+Repository = "https://github.com/SleepyPandas/UofT-MCP"
+
+[project.scripts]
+uoft-mcp = "uoft_mcp.server:main"
+
+[dependency-groups]
+dev = ["pytest>=8,<10", "ruff>=0.11,<1"]
+
+[tool.hatch.build.targets.wheel]
+packages = ["uoft_mcp"]
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+
+[tool.ruff]
+target-version = "py313"
+line-length = 100
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "UP"]

uoft_mcp-0.1.0/tests/conftest.py

@@ -0,0 +1,8 @@
+"""Run async tests with asyncio, matching the stdio server's runtime."""
+
+import pytest
+
+
+@pytest.fixture
+def anyio_backend():
+    return "asyncio"

uoft_mcp-0.1.0/tests/test_stdio.py

@@ -0,0 +1,66 @@
+"""Verify the installed module speaks MCP over actual process pipes, without network calls."""
+
+import asyncio
+import json
+import sys
+from pathlib import Path
+
+import anyio
+import pytest
+from mcp import Client
+from mcp.client.stdio import StdioServerParameters
+
+
+@pytest.mark.anyio
+async def test_sdk_connects_to_stdio_from_another_directory(tmp_path):
+    parameters = StdioServerParameters(
+        command=sys.executable, args=["-m", "uoft_mcp"], cwd=str(tmp_path)
+    )
+    with anyio.fail_after(20):
+        async with Client(parameters, mode="legacy") as client:
+            assert len((await client.list_tools()).tools) == 7
+
+
+@pytest.mark.anyio
+async def test_stdout_contains_only_protocol_json():
+    messages = [
+        {
+            "jsonrpc": "2.0",
+            "id": 1,
+            "method": "initialize",
+            "params": {
+                "protocolVersion": "2025-11-25",
+                "capabilities": {},
+                "clientInfo": {"name": "stdio-test", "version": "1"},
+            },
+        },
+        {"jsonrpc": "2.0", "method": "notifications/initialized"},
+        {"jsonrpc": "2.0", "id": 2, "method": "tools/list", "params": {}},
+    ]
+    process = await asyncio.create_subprocess_exec(
+        sys.executable,
+        "-m",
+        "uoft_mcp",
+        stdin=asyncio.subprocess.PIPE,
+        stdout=asyncio.subprocess.PIPE,
+        stderr=asyncio.subprocess.PIPE,
+        cwd=Path(__file__).resolve().parents[1],
+    )
+    try:
+        for message in messages:
+            process.stdin.write((json.dumps(message) + "\n").encode())
+            await process.stdin.drain()
+            if "id" in message:
+                line = await asyncio.wait_for(process.stdout.readline(), timeout=15)
+                response = json.loads(line)
+                assert response["jsonrpc"] == "2.0"
+                assert response["id"] == message["id"]
+                assert "result" in response
+        process.stdin.close()
+        stdout, stderr = await asyncio.wait_for(process.communicate(), timeout=5)
+        assert not stdout
+        assert process.returncode == 0, stderr.decode()
+    finally:
+        if process.returncode is None:
+            process.kill()
+            await process.wait()