A Claude Code toolkit, a Rust terminal agent, and why 97K developers self-host their photos

The Lens

This is a curated collection of configuration patterns: CLAUDE.md templates, skills, memory structures, security practices, and prompt engineering techniques. Consider it a performance tuning guide for AI-assisted development. MIT licensed, JavaScript. The repo covers agent harness optimization: how to structure your CLAUDE.md for better context, how to write reusable skills, how to set up memory that persists between sessions, and security patterns to prevent prompt injection. It's not a tool you install. It's a reference architecture you adapt to your workflow. Completely free. The repo is MIT licensed and the content is openly available. The ecc.tools website appears to offer additional resources but the core value is in the GitHub repo itself. Solo AI-assisted developers: this is the target audience. If you're using Claude Code, Codex, Cursor, or similar agents and haven't optimized your configuration, there are real gains here. Teams: useful as a starting template for standardizing agent configurations across developers. The catch: this is a knowledge repo, not software. It's only as good as you make it. You need to read, understand, and adapt the patterns to your specific workflow. The star count is massive but the actual value depends on whether you're already using AI coding agents. If you're not, this is irrelevant. If you are, the CLAUDE.md patterns and skill structures alone are worth the read.

The Lens

claurst is an open source terminal coding agent written in Rust, built to replicate the behavior of Claude Code. It reads files, runs commands, searches codebases, and handles git operations from your terminal. The project was built from behavioral specs, not copied source code. The appeal is obvious: Claude Code is a proprietary tool that costs money. claurst gives you a similar workflow for free (assuming you bring your own API key for whatever model you point it at). Being written in Rust means it starts fast and uses less memory than Node.js-based alternatives. The project grew quickly after the Claude Code source leak sparked interest in how these agents work under the hood. For developers who want a terminal coding agent but do not want to pay for Claude Code, this is the most direct alternative. Aider and Continue are more established options with broader model support and larger communities. claurst is newer and less battle-tested, but the Rust foundation and active development are promising. The catch: this is early-stage software riding a wave of hype. The feature set is thinner than Claude Code, the plugin ecosystem does not exist yet, and you are depending on a solo maintainer. If you need reliability today, the established tools are safer bets.

The Lens

CLIProxyAPI wraps existing AI coding CLIs, Gemini CLI, Claude Code, ChatGPT Codex, and others, and exposes them as OpenAI/Gemini/Claude-compatible API endpoints. The pitch is that you get access to models like Gemini 2.5 Pro and GPT-5 through their free CLI tiers, served as a standard API you can plug into any app. Let me be direct: this is a proxy that routes around pricing by using free CLI tools as backends, and exploding because free model access is irresistible. The homepage points to a subscription service at z.ai. The catch: this sits in a gray area. You're wrapping free CLI tools and serving them as APIs, which likely violates the terms of service for most of those CLIs. The sustainability of this approach depends entirely on providers not shutting it down. The MIT license covers the code, but the underlying model access is not yours to redistribute. Use at your own risk.

The Lens

Immich is the self-hosted photo and video management platform that actually competes with the big players. Upload from your phone, browse on the web, search by faces, places, or objects using on-device ML. AGPL v3, TypeScript backend with a Flutter mobile app. Automatic backup from iOS/Android, facial recognition, reverse geocoding, timeline view, shared albums, and a map view. Machine learning runs locally on your server for object detection and face clustering. Self-hosting is free. Docker Compose is the recommended path; you need decent storage and ideally a GPU for faster ML, but CPU works too. Immich offers a paid license for businesses, but the software is fully functional without paying. Solo: perfect. Replace Google Photos for your personal library. Families: shared albums work great. Growing orgs: this isn't designed for enterprise photo management. Large: look at purpose-built DAM (digital asset management) solutions. The catch: Immich is pre-1.0 and the developers explicitly warn against using it as your only backup. The API and schema can change between updates. Keep your originals backed up elsewhere. ML features need real compute: 4GB RAM minimum, more for large libraries.