Hello Codeforces!

I've been working on making Codeforces data more accessible to AI agents — and I'm excited to share something you can start using right away.

Introducing Codeforces MCP Tools — a Model Context Protocol server that turns the Codeforces API into 11 strictly-typed AI tools. Plug it into Claude Desktop, Cursor, VS Code, or any MCP-compatible client, and your AI assistant instantly becomes Codeforces-aware.

What it can do:

• get_user_stats — full profile, rating, rank, solve counts

• get_rating_history — every rating change + latest contest performance

• get_tag_breakdown — how many problems you've solved per tag

• analyze_weak_topics — finds your weakest tags from lowest-rated solves

• get_recommend_problems — smart suggestions based on your rating + weak tags

• get_search_problems — filter problemset by tags / rating range

• get_virtual_contest — custom practice contests excluding solved problems

• get_problem_stats — deep stats for any problem

• get_best_contest_rank — your all-time best contest rank

• get_recent_activity — most recent AC: problem, contest, tags, link

• compare_users — side-by-side comparison of two users

Example prompts your AI can now answer:

"Analyze {userId} and tell me what to practice to reach the next rank."

"Compare {userId1} and {userId2} based on rating history and tags."

"Generate a {N}-problem virtual contest for {userId} rated {minRating}-{maxRating}, excluding already solved problems."

"What are the 3 weakest tags of {userId} and recommend {N} problems to improve them?"

"Show me the rating history and best contest rank of {userId}."

Try it out:

• npm: NPM Package

• GitHub: GitHub Repo

• Smithery: Smithery Server

Watch the setup video: LinkedIn Demo Video

Install in seconds via npm or one-click via Smithery. Full step-by-step setup walkthrough is in the LinkedIn video above.

Feedback, issues, and PRs are very welcome — this is open source and built for the community.

Happy coding!