dsa-tracker-mcp

mcp-name: io.github.ashmitrrr/dsa-tracker-mcp-server

This is an MCP server for tracking your progress through your DSA questions and comes loaded with a default list of NeetCode 150 (any custom DSA problem list can be switched per user) with built-in spaced repetition. Built with FastMCP and SQLite.

Talk to it naturally from Claude: "what should I work on next", "log that I solved Two Sum, confidence 4, took 12 minutes", "how's my progress", "show me my history on Contains Duplicate".

Features

Spaced repetition — problems you struggle with come back sooner, problems you nail come back later. Schedule is based on how you rated each attempt (gave up / struggled / solved) and your confidence (1-5).
List-agnostic — ships with the full NeetCode 150 (150 problems, 18 categories) but can load any custom problem list via a JSON file.
Fuzzy matching — log an attempt with a loosely-typed problem name ("two sum", "contains dupe") and it'll match the right problem.
Stats & streaks — solved counts, per-category breakdown, daily streak, total time spent.

Tools

Tool	Description
`log_attempt`	Log an attempt at a problem (status, confidence 1-5, time spent). Schedules the next review.
`get_next_problem`	Get what to work on next: an overdue review, or the next new problem in order. Optional category filter.
`get_stats`	Overall progress summary — solved counts, per-category breakdown, streak, total time.
`search_problems`	Search/filter problems by name, category, difficulty, or status.
`get_problem_history`	All logged attempts for a given problem, chronological.

Resources

dsa://progress — current progress snapshot
dsa://problem-list — full list of tracked problems

Prompts

daily_review — generates a daily review session based on what's due
explain_pattern(category) — explains the core pattern/approach for a given category

Installation

dsa-tracker-mcp is published on PyPI, no manual cloning or virtual environments needed. The recommended way to run it is with uv, which downloads and runs the package on demand.

Claude Desktop

Add to your claude_desktop_config.json (Settings → Developer → Edit Config):

config.json

{
  "mcpServers": {
    "dsa-tracker": {
      "command": "uvx",
      "args": ["dsa-tracker-mcp"]
    }
  }
}

Restart Claude Desktop completely after saving.

Alternative: pip install

If you'd rather install it directly:

Terminal

pip install dsa-tracker-mcp

Then point your config at the installed console script:

config.json

{
  "mcpServers": {
    "dsa-tracker": {
      "command": "dsa-tracker-mcp"
    }
  }
}

From source

bash

git clone https://github.com/ashmitrrr/dsa-tracker-mcp-server.git
cd dsa-tracker-mcp-server
python3 -m venv .venv
source .venv/bin/activate   # on Windows: .venv\Scripts\activate
pip install -e .

Environment variables (optional)

Variable	Default	Description
`DSA_TRACKER_DB`	`~/.dsa_tracker_mcp/progress.db`	Path to the SQLite database
`DSA_TRACKER_PROBLEMS_FILE`	(none, uses built-in NeetCode 150)	Path to a JSON file with a custom problem list

To set these with uvx, add an env block to your config:

config.json

{
  "mcpServers": {
    "dsa-tracker": {
      "command": "uvx",
      "args": ["dsa-tracker-mcp"],
      "env": {
        "DSA_TRACKER_PROBLEMS_FILE": "/absolute/path/to/my-problems.json"
      }
    }
  }
}

Custom problem list format

config.json

[
  {
    "name": "Two Sum",
    "category": "Arrays & Hashing",
    "difficulty": "Easy",
    "url": "https://leetcode.com/problems/two-sum/"
  }
]

url and difficulty are optional and will be auto-filled where possible. order_index is assigned automatically based on list order.

Spaced repetition logic

Outcome	Next review
Gave up	1 day
Struggled, confidence ≤ 2	2 days
Struggled, confidence ≥ 3	4 days
Solved, confidence ≤ 3	7 days
Solved, confidence ≥ 4	21 days

get_next_problem prioritizes overdue reviews before suggesting new problems.

Example prompts

"What should I work on next?"
"I just solved Valid Anagram, confidence 5, took 6 minutes, log it"
"How's my progress on Trees?"
"Show me my history on Two Sum"
"Give me a daily review"

License

MIT — see LICENSE.