About AtCoder MCP Server
AtCoder MCP Server lets users find past AtCoder problems and retrieve their public statements, locally or through a hosted HTTP endpoint. Its read-only tools do not run or submit solutions and require no judge account credentials.
Key features
- 01Search problems by URL or task ID
- 02Fetch public problem statements
- 03Check server health and capabilities
- 04Hosted HTTP and local stdio options
Use cases
- 01Find a past problem by contest and task ID
- 02Retrieve a problem statement and samples
- 03Check hosted server readiness
- 04Practice with past AtCoder problems
MCP support
The MCP server exposes read-only AtCoder problem search and fetch tools. It can be accessed through hosted HTTP or run locally over stdio; no API key or judge account is required.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://api.ksrnyx.top/oj-mcp/atcoder/mcpclaude mcp add --transport http ketherworks-atcoder https://api.ksrnyx.top/oj-mcp/atcoder/mcp{"mcpServers":{"ketherworks-atcoder":{"type":"http","url":"https://api.ksrnyx.top/oj-mcp/atcoder/mcp"}}}Open in Cursor{"servers":{"ketherworks-atcoder":{"type":"http","url":"https://api.ksrnyx.top/oj-mcp/atcoder/mcp"}}}Open in VS Code{"mcpServers":{"ketherworks-atcoder":{"type":"http","url":"https://api.ksrnyx.top/oj-mcp/atcoder/mcp"}}}{"mcpServers":{"ketherworks-atcoder":{"serverUrl":"https://api.ksrnyx.top/oj-mcp/atcoder/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
4This tool takes no parameters.
Details
From the project READMEAtCoder MCP Server
Find and read past AtCoder problems through MCP, locally or over hosted HTTP.
Quick Start
Add the hosted server to your MCP configuration:
{
"servers": {
"atcoder": {
"type": "http",
"url": "https://api.ksrnyx.top/oj-mcp/atcoder/mcp"
}
}
}
The endpoint accepts anonymous read requests. It does not need an API key, Cookie, or judge account.
Then ask:
Fetch AtCoder problem abc086_a and show its samples.
What It Can Do
| Tool | Purpose |
|---|---|
oj_capabilities | Report the available AtCoder read operations and transport. |
oj_health | Report process readiness and the latest upstream read. |
oj_fetch_problem | Fetch one past problem by canonical URL or contest/task id. |
oj_search_problems | Resolve an exact task URL or contest/task identifier. |
Run Locally
Requires Node.js 22 or newer.
npm ci
npm run check
npm run build
node packages/atcoder/dist/index.js
MCP client configuration from a source checkout:
{
"servers": {
"atcoder": {
"type": "stdio",
"command": "node",
"args": ["/absolute/path/packages/atcoder/dist/index.js"]
}
}
}
Tagged GitHub releases attach a standalone npm tarball.
Availability
- Hosted MCP:
https://api.ksrnyx.top/oj-mcp/atcoder/mcp - Official MCP Registry:
io.github.ketherworks/atcoder, described byserver.json - Health status: see the repository homepage or call
oj_health
Source and Safety
This standalone release is generated from the reviewed Competitive Programming MCP source. The release package bundles the shared OJ contract implementation, so its runtime does not depend on unpublished workspace packages.
Platform Rules
Do not use this server during ongoing ABC, ARC, or AGC contests. AtCoder's current generative-AI rules generally prohibit that use. This project is intended for practice with past problems. It is unofficial and is not affiliated with or endorsed by AtCoder Inc. Problem content remains subject to the AtCoder Terms of Use.
The server exposes no run or submit tool. It accepts no judge account credentials. See SECURITY.md for the security boundary and PROVENANCE.md for the canonical source revision.
Development
npm ci
npm run check
npm run pack:check
npm run deploy:dry
License
Adapter source code is MIT licensed. Judge problem content and trademarks are not relicensed by this repository.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.