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., "@algo-coach-mcp Pick a random easy problem on hash tables"
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.
Interactive algorithm coach MCP server with 3000+ LeetCode problems, bilingual descriptions (Chinese/English), local code testing, and real-world engineering case mapping.
3000+ Problems — Full LeetCode free problem database with bilingual descriptions
12 Topic Categories — Array, Linked List, Hash Table, String, Two Pointers, Stack/Queue, Binary Tree, Backtracking, Greedy, DP, Monotonic Stack, Graph
Local Code Execution — Run and test your Python solutions locally, no online judge needed
Progressive Hints — 4-level hint system: direction -> approach -> pseudocode -> full solution
Real-World Cases — See how algorithms apply in production systems (Redis, Kafka, React, etc.)
3 Practice Modes — Student (guided), Interview (timed), Engineering (system design focus)
LeetCode Sync — Built-in script to fetch and update problems from LeetCode API
Related MCP server: Swarm Orchestrator
One command setup — no local installation needed:
claude mcp add --transport stdio algo-coach -- npx -y --registry https://registry.npmjs.org/ algo-coach-mcp@latestRestart Claude Code, then start practicing.
Add to your MCP configuration:
{
"mcpServers": {
"algo-coach": {
"type": "stdio",
"command": "npx",
"args": ["-y", "--registry", "https://registry.npmjs.org/", "algo-coach-mcp@latest"]
}
}
}Tool | Description |
| Pick a random problem by topic/difficulty |
| Get solution code and key points |
| Get theoretical fundamentals for a topic |
| Real-world engineering applications of an algorithm |
| Generate boundary test cases for a problem |
| Execute Python code against tests locally |
| Get the learning progression |
# | Topic | Description |
1 | Array (数组) | Binary search, two pointers, sliding window |
2 | Linked List (链表) | Reversal, cycle detection, merge |
3 | Hash Table (哈希表) | Lookup, grouping, counting |
4 | String (字符串) | Matching, parsing, manipulation |
5 | Two Pointers (双指针) | Fast-slow, left-right, sliding window |
6 | Stack & Queue (栈与队列) | Monotonic queue, expression parsing |
7 | Binary Tree (二叉树) | Traversal, construction, BST |
8 | Backtracking (回溯) | Permutations, combinations, subsets |
9 | Greedy (贪心) | Interval scheduling, optimization |
10 | Dynamic Programming (动态规划) | Knapsack, subsequence, state machines |
11 | Monotonic Stack (单调栈) | Next greater element, histogram |
12 | Graph (图论) | BFS, DFS, union-find, topological sort |
Fetch all free problems from LeetCode with bilingual descriptions:
npm run sync # Full sync (~3000 problems, ~50 min)
npm run sync -- --limit 100 # Sync first 100 problems
npm run sync:resume # Resume interrupted syncFeatures: checkpoint/resume, rate limiting (2 req/s), retry logic, bilingual (CN + EN).
npm install
npm run dev # Run with tsx (hot reload)
npm run sync # Sync problems from LeetCode
npm run build # Build for production
npm test # Run testssrc/
├── index.ts # MCP server entry (stdio transport)
├── paths.ts # Package root resolution
├── types.ts # Shared type definitions
├── content/ # Content indexing and parsing
├── sync/ # LeetCode API sync pipeline
├── testgen/ # Test case generation
├── executor/ # Python subprocess runner
├── cases/ # Real-world case loader
├── tools/ # MCP tool implementations
└── resources/ # MCP resource handlersMIT
Generate test cases (including edge cases) for a problem
Inputs: slug (required).
Get real-world engineering application cases for an algorithm
Inputs: algorithm (required).
Get the solution code and key points for a problem
Inputs: slug (required), language.
Get the theory/fundamentals article for a topic
Inputs: topic (required).
Get the ordered topic roadmap for learning progression
Inputs: no named parameters.
Pick a random problem by topic and/or difficulty
Inputs: topic, difficulty.
Run user-submitted Python code against test cases locally
Inputs: code (required), testCases (required), functionName (required), language, +1 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 7 tools
Each tool clearly targets a distinct action: picking problems, retrieving solutions, theory, real-world cases, test generation, code execution, and roadmap. No two tools overlap in purpose, making misselection unlikely.
All tool names follow a consistent verb_noun snake_case pattern (pick_problem, get_solution, generate_test_cases, etc.). The naming is predictable and clearly indicates each tool's function.
Seven tools is a well-scoped set for an algorithm coaching server. Each tool addresses a distinct part of the learning workflow without redundancy or bloat, fitting the typical ideal range.
The tool surface covers the essential learning loop: selecting problems, accessing theory/solutions, generating tests, running user code, and following a roadmap. There are no obvious gaps or dead ends for the stated coaching purpose.
Remote MCP learning coach for coding agents.
Educational MCP server with 17 math/stats tools, visualizations, and persistent workspace
Remote MCP server for deterministic educational practice-assessment score conversions.
Persistent memory and cross-session learning for AI coding assistants (hosted remote MCP).
