2 Open-Source Skills Projects: Obsidian Skills & Agent-Skills-for-Context-Engineering

Obsidian Skills: Open-Source AI Guide for Obsidian Notes

Many readers who follow our content are likely already using Obsidian—a knowledge management and note-taking software based on local Markdown files.

Obsidian-skills is an open-source project personally launched by Steph Ango, CEO of Obsidian. It is specifically designed to teach Claude Code how to correctly understand, write, and manage Obsidian note files.

In less than a week, the project has already garnered 5K stars on GitHub.

《2 Open-Source Skills Projects: Obsidian Skills & Agent-Skills-for-Context-Engineering》

Numerous users leverage Claude Code to directly interact with their local Obsidian vaults. However, general-purpose AI lacks a full understanding of Obsidian-specific syntax, such as bidirectional link formatting, Canvas JSON structures, and more. This project equips AI with specialized Obsidian expertise.

It includes a set of skills that, when integrated with Claude Code, enable the AI to master Obsidian-related operations:

① Mastery of Obsidian Markdown Syntax

Covers syntax for bidirectional links, embeds, callouts, frontmatter, and other Obsidian-specific formatting—all clearly documented in the skill files.

② Canvas Creation Capabilities

Obsidian Canvas is essentially a complex JSON file; this skill teaches AI to properly structure and generate Canvas content.

③ Obsidian Bases Support

Tailored for Obsidian’s database/table functionality, it instructs AI on writing filters, formulas, aggregations, and function calls—preventing the AI from inventing non-existent functions.

Open-Source Link: https://github.com/kepano/obsidian-skills

Context Engineering Skills Library

After the rise of Manus, attention has gradually shifted from prompt engineering to context engineering. The core focus of the open-source project “Agent-Skills-for-Context-Engineering” lies in this very domain: how to scientifically manage large language models’ (LLMs) context windows to maximize Agent performance.

Boasting a comprehensive library of skills for building production-grade AI Agent systems, the project garnered over 6,000 stars in just two weeks.

《2 Open-Source Skills Projects: Obsidian Skills & Agent-Skills-for-Context-Engineering》

These skills revolve around the holistic planning of all information fed into the model—including system prompts, tool definitions, retrieved documents, conversation history, and tool outputs.

As context length increases, models face issues like “lost-in-the-middle,” attention scarcity, and performance degradation. The skills provided by this project aim to help developers identify the minimal set of high-signal tokens to achieve optimal results.

Agents only load skill names and descriptions at startup; full content is loaded only when specific tasks are activated. This “progressive disclosure” approach saves valuable context space.

The project categorizes skills into four areas, covering everything from foundational theory to advanced architectural design:

① Foundational Skills

  • Context Fundamentals: Understand the anatomy of context and its significance.
  • Context Degradation: Identify patterns of context failure (e.g., contamination, interference, conflicts).
  • Context Compression: Design compression strategies for long conversations.

② Architectural Skills

  • Multi-Agent Patterns: Master orchestrator, peer-to-peer (P2P), and hierarchical multi-agent architectures.
  • Memory Systems: Design short-term, long-term, and graph-based memory architectures.
  • Tool Design: Build tools (e.g., MCP tools) that Agents can use effectively.
  • Filesystem Context: Leverage file systems for dynamic context discovery and state persistence.

③ Operational Skills

  • Optimization: Context compression, masking, and caching strategies.
  • Evaluation: Build evaluation frameworks for Agent systems.
  • Advanced Evaluation: Master LLM-as-a-Judge techniques, including direct scoring, pairwise comparison, and bias mitigation.

④ Cognitive Architecture Skills

  • BDI Mental States: Convert external context into an Agent’s mental states based on the BDI (Belief-Desire-Intention) model to enable more rational reasoning.

How to Use

The project offers multiple usage methods to adapt to different development environments:

For Claude Code Users

Treat it as a Claude Code plugin marketplace. Install specific skill packages directly via commands:

plaintext

/plugin marketplace add muratcankoylan/Agent-Skills-for-Context-Engineering  
/plugin install agent-architecture@context-engineering-marketplace

After installation, Claude will automatically activate relevant skills based on your task context.

For Cursor Users

Copy the content of SKILL.md into your project’s .rules file or a designated folder. It will serve as a background knowledge base for the AI coding assistant, helping it better understand Agent construction.

Open-Source Link: https://github.com/muratcankoylan/Agent-Skills-for-Context-Engineering