Online Judge MCP Server
This MCP server acts as an interface for an Online Judge platform, allowing AI agents to interact with coding problems and submissions.
List Problems (
list_problems): Retrieve a paginated list of available coding problems, with configurable offset and limit parameters.List My Submissions (
list_my_submissions): Fetch the authenticated user's recent submission history, also supporting pagination.Get Problem Details (
get_problem_details): Retrieve full details of a specific problem by its display ID (e.g.,PR-114-1-31), including descriptions, sample inputs/outputs, limits, and rules.Submit Code (
submit_code): Submit source code (C, C++, Python3, etc.) to a specific problem using its internal system ID.Get Submission Status (
get_submission_status): Poll and check the result/status of a previously submitted solution using its submission ID.
Authentication uses username/password credentials, and the server is compatible with MCP clients such as Claude Desktop, Cursor, Cline, and OpenCode IDE.
Click on "Deploy 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., "@Online Judge MCP Server List the latest problems and show me the details for problem 1001"
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.
Online Judge MCP Server
This is a Model Context Protocol (MCP) server that provides AI agents with tools to interact with an Online Judge API.
Features
Provides the following tools:
list_problems(offset, limit): List available problems on the OJ.list_my_submissions(offset, limit): List recent submissions made by the authenticated user.get_problem_details(problem_id): Fetch detailed descriptions, samples, limits, and rules of a problem.submit_code(problem_id, code, language): Submit source code to a specific problem.get_submission_status(submission_id): Check the polling status of a code submission.
Related MCP server: MCP Chef
Quickstart
Using uvx (Recommended)
The easiest way to run this server is using uvx, which automatically downloads the correct Python version, creates an ephemeral environment, and runs the tool seamlessly:
set OJ_URL="http://localhost:8000" # Replace with actual OJ URL
set OJ_USERNAME="your-username"
set OJ_PASSWORD="your-password"
uvx oj-mcp-serverUsing pip or pipx
Alternatively, you can install the package directly into your Python environment:
pip install oj-mcp-server
oj-mcp-serverUsage with MCP Clients
To use this server, add it to your MCP client's configuration (e.g., Claude Desktop, Cursor, Cline).
Claude Desktop
Add the following to your claude_desktop_config.json:
{
"mcpServers": {
"oj-mcp-server": {
"command": "uvx",
"args": ["oj-mcp-server"],
"env": {
"OJ_URL": "http://localhost:8000",
"OJ_USERNAME": "your-username",
"OJ_PASSWORD": "your-password"
}
}
}
}Note: If you installed globally via pip, simply change "command" to "oj-mcp-server" and "args" to [].
Cursor & Cline
In the MCP settings panel, add a new server:
Type: command
Name:
oj-mcp-serverCommand:
uvx oj-mcp-server(oroj-mcp-serverif installed via pip)Environment: Set
OJ_URL,OJ_USERNAME, andOJ_PASSWORD
OpenCode
Add this configured server inside your opencode.json (under the mcp key):
{
"mcp": {
"oj-mcp-server": {
"type": "local",
"command": ["uvx", "oj-mcp-server"],
"environment": {
"OJ_URL": "http://localhost:8000",
"OJ_USERNAME": "your-username",
"OJ_PASSWORD": "your-password"
}
}
}
}Note: If using pip, change the command array to ["oj-mcp-server"].
Available Tools
5 toolsget_problem_detailsB
Fetch the detailed information of a specific problem.
problem_id: The external ID/display ID of the problem (e.g., "PR-114-1-31").
Inputs: problem_id (required). Returns: result
get_submission_statusB
Get the status and result of a previously submitted code.
submission_id: The ID returned when submitting the code.
Inputs: submission_id (required). Returns: result
list_my_submissionsC
Fetch the list of latest submissions made by the authenticated user. Returns a JSON string containing the submission histories.
Inputs: offset, limit. Returns: result
list_problemsC
Fetch the list of problems from the Online Judge. Returns a JSON string containing total count and results.
Inputs: offset, limit. Returns: result
submit_codeB
Submit code to the Online Judge for a specific problem.
problem_id: the internal system ID of the problem.
language: string such as "C", "C++", "Python3".
code: the source code to submit.
Inputs: problem_id (required), code (required), language (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool has a clearly distinct purpose with no ambiguity: get_problem_details retrieves specific problem info, get_submission_status checks submission results, list_my_submissions shows user's submission history, list_problems lists available problems, and submit_code submits new code. The boundaries are well-defined and non-overlapping.
All tools follow a consistent verb_noun pattern using snake_case: get_problem_details, get_submission_status, list_my_submissions, list_problems, and submit_code. The naming is predictable and readable throughout the set.
With 5 tools, this server is well-scoped for an online judge domain. Each tool earns its place by covering core workflows: browsing problems, getting details, submitting code, and checking results. The count is neither too thin nor excessive for the purpose.
The toolset covers the essential CRUD-like operations for an online judge: list and get for problems, create (submit) for submissions, and get for submission status. A minor gap is the lack of a tool to delete or cancel submissions, but agents can work around this, and core workflows are well-supported.
Maintenance
Related MCP Connectors
Search Luogu problems, fetch statements, explore problem sets and get practice recommendations.
Search Codeforces problems and inspect public problem metadata through the official Codeforces API.
Task and bug tracker for AI coding agents: 75 tools to file, triage and update work.
Search AtCoder problems and fetch public problem statements through MCP.
Related MCP Servers
- AlicenseCqualityDmaintenanceEnables code execution and compilation through the Judge0 CE API, supporting multiple programming languages with submission management, status checking, and language configuration retrieval.9MIT
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to participate in CodeChef contests by fetching problems, generating and testing solutions in a secure sandbox, and submitting answers.-
- AlicenseAqualityBmaintenanceConnects AI assistants to the Repovive competitive-programming platform, enabling them to browse problems and contests, judge and submit code for real verdicts, track ratings, communicate with users, and author verified problems and rounds.102MIT
- FlicenseNot gradedqualityAmaintenanceEnables automated access to problems, contests, solutions, and public or authenticated user data across global and China platforms, with tools for searching problems, retrieving profiles and submissions, running code, submitting solutions, and managing notes.-