1. AgentsScore87

    Intent CLI: Semantic Git for AI Coding Sessions

    Intent CLI introduces a semantic history layer for AI coding agents, storing project goals, verified checkpoints, and decisions in plain JSON. This aims to help agents retain context across sessions, preventing the need for repeated explanations.

    Intent CLI: Semantic Git for the vibe coding era, available on GitHub. Full analysis
  2. AI CodingScore87

    Blender Agent Skills for AI Code Assistants

    This repository provides professional Blender 5.2 workflows as agent skills for AI code assistants like Claude Code, Codex, Gemini CLI, Copilot, and Cursor. It covers a range of 3D tasks from modeling to export.

    Source: EL4CTEO/blender-skills on GitHub Full analysis
  3. AgentsScore89

    Expo Exact: Cross-Platform App Development with AI Agents

    Expo Exact is a Rust-based project enabling developers to write an application once, which then runs natively across web, macOS, iOS, and Linux. An AI agent can build, run, and test these applications on all supported platforms.

    Source: https://github.com/expo/exact Full analysis
  4. AgentsScore88

    Strands Box: Open Source AI Agent Sandbox

    Strands Box is an open-source sandbox for AI agents, written in Rust. It restricts agent execution, file access, and network reach using OS isolation and semantic policies defined in Dogwood. The system supports macOS on Apple silicon and Linux ARM64, with experimental x86-64 support.

    Source: Strands Box GitHub Repository (https://github.com/strands-agents/box) Full analysis
  5. AI CodingScore87

    TokenSafe v2.9.0 Optimizes AI Agent Costs

    TokenSafe has released version 2.9.0, a Python-based tool designed to significantly reduce AI agent operational costs by up to 99%. The tool achieves this by optimizing code reading, caching prompts across interactions, and preventing secrets from being exposed in git.

    Source: tokensafe-dev/tokensafe on GitHub Full analysis