KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#typescript #ai-agents #cli #claude-code #codex #developer-tools #dsh-plugin #deepseek-harness #windows #open-source #agent-skills #ai

grindx

★ 40 Open GitHub ↗

A distraction free DSA practice in your terminal

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# grindx

Distraction-free DSA practice in your terminal.

![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-blue)
![License: MIT](https://img.shields.io/badge/license-MIT-green)

![grindx demo](https://raw.githubusercontent.com/xghostient/grindx/main/demo-grindx.gif)

## Why grindx?

- **Local-first** — problem browsing, progress, and solutions stay local; AI review and testcase bundle downloads are optional
- **Terminal-native** — practice DSA without leaving your terminal or opening a browser
- **Multiple sheets** — Striver A2Z (316), Blind 75, NeetCode 150, Grind 75 built-in
- **5 languages** — Python, Go, C++, Java, JavaScript — switch on the fly with `Ctrl+L`
- **Progress tracking** — solved/in-progress/not-started states, streaks, best times
- **Bookmarks & filters** — filter by difficulty (Easy/Medium/Hard) or bookmarked problems

## Install

```bash
pip install grindx
```

Or with [pipx](https://pipx.pypa.io/) (recommended for CLI tools):

```bash
pipx install grindx
```

Or run from source:

```bash
git clone https://github.com/xghostient/grindx.git
cd grindx
python3 -m venv .venv && source .venv/bin/activate
pip install -e .
grindx
```

### Local testcase bundle

PyPI installs keep the package lean and do not ship the full bundled testcase corpus. To enable local `Ctrl+R` test runs, fetch the external testcase bundle once:

```bash
grindx --fetch-testcases
```

By default this pulls from:

```text
https://github.com/grindxhq/dsa-catalog/releases/latest/download/manifest.json
```

Override the manifest source if needed:

```bash
grindx --fetch-testcases --testcase-manifest-url https://example.com/manifest.json
# or
export GRINDX_TESTCASE_MANIFEST_URL=https://example.com/manifest.json
```

## Usage

```bash
grindx            # CLI entry point
python -m grindx  # or as a module
```

### Navigation

| Key | Action |
|-----|--------|
| `↑` `↓` | Navigate topics / problems |
| `←` `→` | Switch between topic and problem panes |
| `Enter` | Select topic or open problem |
| `Esc` | Go back |
| `q` | Quit |

### Filters

| Key | Filter |
|-----|--------|
| `a` | All problems |
| `e` | Easy |
| `m` | Medium |
| `h` | Hard |
| `b` | Bookmarked |
| `s` | Stats dashboard |

### Solve Screen

| Key | Action |
|-----|--------|
| `Ctrl+S` | Save code |
| `Ctrl+D` | Toggle solved |
| `Ctrl+E` | AI review |
| `Ctrl+L` | Cycle language (Python → Go → C++ → Java → JS) |
| `Ctrl+B` | Toggle bookmark |
| `Ctrl+T` | Pause / resume timer |
| `Ctrl+R` | Run tests |
| `Ctrl+Shift+R` | Reset timer |
| `Ctrl+Shift+C` | Copy selection to clipboard |
| `Ctrl+Shift+V` | Paste from clipboard |
| `Alt+↑` / `Alt+↓` | Move line up / down |
| `Alt+Shift+↓` | Duplicate line |
| `Esc` | Save & go back |

## Features

**Split-pane editor** — problem description on the left, code editor with syntax highlighting on the right.

**Auto-timer** — starts when you open a problem, tracks your best solve time.

**Three-state tracking** — each problem shows as not started (○), in progress (◐), or solved (✓).

**Stats dashboard** — overall progress, per-difficulty breakdown, per-topic progress bars, current streak, and top 10 best times.

**Sheet-agnostic** — built-in sheets live inside the package (`grindx/sheets/`). Format:

```json
{
  "Topic Name": ["problem-name-1", "problem-name-2"],
  "Another Topic": ["problem-name-3"]
}
```

**Progress safety** — automatic backups with corruption recovery. Progress, solutions, backups, and downloaded testcase bundles are stored in `~/.grindx/` so they persist across installs and upgrades.

## AI Review (optional)

Press `Ctrl+E` on the solve screen to get AI-powered feedback on your solution — correctness, edge cases, complexity analysis, and a pass/fail verdict.

### Setup

Set two environment variables:

```bash
export GRINDX_AI_PROVIDER=groq              # or ollama, anthropic, openai, bedrock
export GRINDX_AI_MODEL=llama-3.3-70b-versatile  # optional, sensible defaults per provider
export GRINDX_AI_KEY=gsk_...                # not needed for ollama or bedrock
export GRINDX_AI_URL=https://custom.api/v1  # optional, auto-detected per provider
```

Or create `~/.grindx.toml`:

```toml
[ai]
provider = "groq"
model = "llama-3.3-70b-versatile"
api_key = "gsk_..."
```

### Supported Providers

| Provider | API Key | Default Model | Notes |
|----------|---------|---------------|-------|
| `ollama` | No | llama3 | Local, free, no network |
| `groq` | Yes | llama-3.3-70b-versatile | Fast, free tier available |
| `anthropic` | Yes | claude-sonnet-4-20250514 | Claude |
| `openai` | Yes | gpt-4o | GPT |
| `bedrock` | No (IAM/SSO) | claude-sonnet (Bedrock) | AWS Bedrock, supports inference profile ARNs |

Any OpenAI-compatible API works — set provider to `openai` and add `base_url`.

### AWS Bedrock

For corporate environments using AWS Bedrock with IAM or SSO authentication:

```toml
[ai]
provider = "bedrock"
model = "anthropic.claude-sonnet-4-20250514-v1:0"
aws_profile = "my-profile"   # optional, uses default credential chain if omitted
aws_region = "us-west-2"     # optional, uses boto3 default if omitted
```

The `model` field accepts both native Bedrock model IDs (e.g. `anthropic.claude-sonnet-4-20250514-v1:0`) and inference profile ARNs (e.g. `arn:aws:bedrock:us-west-2:123456:application-inference-profile/abc123`).

Requires `boto3` (`pip install boto3`) and valid AWS credentials (SSO session, env vars, instance profile, etc.).

## Custom Sheets & Problems

You can add your own problem sheets and problems alongside the built-in ones.

### Custom sheet

Create a JSON file in `~/.grindx/sheets/`:

```json
// ~/.grindx/sheets/my-list.json
{
  "Arrays": ["two-sum", "best-time-to-buy-and-sell-stock", "my-custom-problem"],
  "Strings": ["valid-anagram", "another-custom-problem"]
}
```

You can mix built-in problem IDs (like `two-sum`) with your own custom problem IDs. Run `grindx --list-problems` to see all available built-in problem IDs.

### Custom problems

Create a folder per problem under `~/.grindx/problems/`:

```json
// ~/.grindx/problems/my-custom-problem/problem.json
{
  "id": "my-custom-problem",
  "name": "My Custom Problem",
  "difficulty": "Medium",
  "category": "Arrays",
  "description": "Given an array of integers, find ...",
  "examples": [
    {"input": "nums = [1, 2, 3]", "output": "6"},
    {"input": "nums = []", "output": "0"}
  ],
  "constraints": "1 <= nums.length <= 10^4",
  "python_template": "def solve(nums):\n    pass\n",
  "go_template": "func solve(nums []int) int {\n\n}\n",
  "cpp_template": "int solve(vector<int>& nums) {\n\n}\n",
  "java_template": "class Solution {\n    public int solve(int[] nums) {\n\n    }\n}\n",
  "js_template": "function solve(nums) {\n\n}\n"
}
```

Optional local judge assets for a custom problem:

```text
~/.grindx/problems/my-custom-problem/testcases.json
~/.grindx/problems/my-custom-problem/judges/python.py
~/.grindx/problems/my-custom-problem/judges/go.go
~/.grindx/problems/my-custom-problem/judges/cpp.cpp
~/.grindx/problems/my-custom-problem/judges/java.java
~/.grindx/problems/my-custom-problem/judges/javascript.js
```

You only need to include the template keys and local judge assets for the languages you care about.

Legacy `~/.grindx/problems/custom.json` is still loaded for compatibility, but new custom problems should use the folder-based format above.

Custom sheets show up on the dashboard alongside the built-in ones.

## Built with

- [Textual](https://github.com/Textualize/textual) — TUI framework
- [tree-sitter](https://tree-sitter.github.io/) — syntax highlighting

## License

MIT

grindx

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README