AI Open Source Projects · August 18, 2026

Curated selection of the most impactful open-source AI projects this week.

1. affaan-m/ECC ⭐ 240.7k

What it does: ECC (Agent Harness Performance Optimization System) is a performance-optimization framework for AI agent development harnesses, focused on improving the stability and efficiency of agent tools like Claude Code, Codex, and Opencode. It blends skills, instincts, memory, security, and research-first development into a cohesive toolkit for building more robust agent workflows.

Key features:

Deep performance profiling and optimization for agent harnesses
Built-in memory and security mechanisms
Support for Claude Code, Codex, Opencode, and more
Research-driven development methodology

GitHub: https://github.com/affaan-m/ECC

2. NousResearch/hermes-agent ⭐ 232.0k

What it does: Hermes Agent is an open-source AI agent framework by Nous Research — “the agent that grows with you.” It emphasizes extensibility and personalization, letting users shape their agent’s capabilities through skills, persistent memory, and custom tools. The project enjoys an extremely active community and continues to ship updates this week.

Key features:

Modular skill system with persistent memory
Multi-model backend support (Anthropic, OpenAI, and more)
Custom tools and workflow orchestration
Active community and rapid iteration

GitHub: https://github.com/NousResearch/hermes-agent

3. n8n-io/n8n ⭐ 201.0k

What it does: n8n is a fair-code workflow automation platform with native AI capabilities, combining visual building with custom code and self-hosted deployment across 400+ integrations. Developers can assemble automation flows that chain LLM calls, data transformations, and multi-system actions — making it a de facto standard for AI workflow automation.

Key features:

Visual workflow canvas plus custom code nodes
Native AI nodes (LLM, vector stores, agents)
400+ third-party integrations
Self-hosted or cloud deployment

GitHub: https://github.com/n8n-io/n8n

4. firecrawl/firecrawl ⭐ 168.6k

What it does: Firecrawl is a web data extraction engine built for the AI era, offering “the context API” that turns any webpage into clean, LLM-friendly data. Point it at a URL and get structured Markdown, JSON, or screenshots — with support for whole-site crawling, search, and interactive scraping. It is one of the most popular open-source ways for AI agents to fetch live web data.

Key features:

Webpage-to-Markdown/JSON structured output
Whole-site crawling, search, and scheduled scraping
Built-in anti-bot and rendering handling
Native TypeScript with a simple API

GitHub: https://github.com/firecrawl/firecrawl

5. Graphify-Labs/graphify ⭐ 107.5k

What it does: Graphify turns any codebase — including its docs, SQL schemas, configs, and PDFs — into a queryable knowledge graph. It ships /graphify skills for Claude Code, Cursor, Codex, and similar tools, letting developers retrieve project structure and trace dependencies in natural language. Launched only months ago, it has already crossed the hundred-thousand-star mark.

Key features: