Erold is an AI-native project management platform built specifically for developers and AI agents to collaborate. It features MCP server integration, REST API, CLI tool, and a knowledge base that gives AI persistent memory across sessions.

How does Erold integrate with Claude Code?

Erold provides a native MCP (Model Context Protocol) server with 18 tools that Claude Code can use directly. This allows Claude to create tasks, update progress, access project knowledge, and manage your projects without any copy-paste.

Yes! Erold offers a free tier with 1 project and unlimited tasks. Pro is $5/month with pay-as-you-go pricing for extra resources. Only pay for what you use beyond included allocations.

What AI tools does Erold support?

Erold works with any AI tool that supports MCP or REST APIs, including Claude Code, Claude Desktop, Cursor, Windsurf, GitHub Copilot, and custom AI agents. The API-first design ensures compatibility with current and future AI tools.

What is MCP (Model Context Protocol)?

MCP (Model Context Protocol) is a standard protocol that allows AI assistants to interact with external tools and services. Erold's MCP server provides 18 tools for task management, knowledge base access, and project operations that AI assistants can use directly.

How is Erold different from Jira, Linear, or Asana?

Erold is built API-first for AI agents, not humans-first with APIs added later. Traditional tools require copy-paste to share context with AI. Erold lets AI assistants read tasks, update progress, and access project knowledge directly through MCP, API, or CLI.

What is the Knowledge Base feature?

The Knowledge Base is persistent memory for AI assistants. Store project context, coding standards, architecture decisions, and any information AI should remember across sessions. AI can read and write knowledge entries, making it smarter about your specific project over time.

Does Erold have a CLI tool?

Yes! The Yet CLI (npm install -g @erold/cli) provides full task management from your terminal. Create tasks, list projects, mark items done, and integrate with git workflows. Perfect for developers who prefer command-line tools.

When should I use NumPy Patterns?

Numerical computing in Python. Array/matrix operations. Scientific computing. ML preprocessing

When should I NOT use NumPy Patterns?

GPU computing (use CuPy/JAX). Distributed arrays (use Dask). Symbolic math (use SymPy)

Data python recommended

NumPy Patterns

NumPy patterns covering broadcasting, vectorization, memory layout, universal functions, array creation, and performance optimization.

Difficulty: intermediate
Read time: 1 min read
Version: v1.0.0
Confidence: established
Last updated: Updated Jan 9, 2026

Edit on GitHub Report Issue

Quick Reference

NumPy: Vectorize everything (no Python loops). Use broadcasting for shape-compatible operations. Pre-allocate arrays (np.empty, np.zeros). Use out= parameter to avoid temporaries. Prefer C-contiguous arrays. Use appropriate dtypes (float32 vs float64). Boolean indexing for filtering. np.einsum for complex operations.

Use When

Numerical computing in Python
Array/matrix operations
Scientific computing
ML preprocessing

Skip When

GPU computing (use CuPy/JAX)
Distributed arrays (use Dask)
Symbolic math (use SymPy)

NumPy Patterns

NumPy patterns covering broadcasting, vectorization, memory layout, universal functions, array creation, and performance optimization.

Discussion

Data Pipelines recommended Pandas Patterns recommended

NumPy Patterns

Quick Reference

Use When

Skip When

NumPy Patterns

Discussion

Strictly Necessary

Analytics

Functional