Agentic Coding

The agent-skills repository provides 24 production-grade engineering skills designed specifically for AI coding agents like Claude Code and Cursor. It structures the entire development lifecycle—from defining requirements through shipping—into eight slash commands (/spec, /plan, /build, /test, /review, /webperf, /code-simplify, /ship) that automatically activate relevant skills based on the task. Each skill encodes workflows, quality gates, and best practices from senior engineers, including PRD-driven spec work, test-driven development, code review discipline, and shipping patterns. The pack emphasizes anti-rationalization tables that counter common shortcuts (like "I'll add tests later"), verification gates at every stage, and enforcement of practices like atomic commits, feature flags, and staged rollouts. It installs directly into Claude Code via the marketplace, Cursor by copying SKILL.md files into .cursor/rules/, and other coding environments, making structured workflows and engineering discipline consistently available across every coding session.

Gentle-AI is an ecosystem configurator that enhances AI coding agents like Claude Code and Cursor with persistent memory (Engram), Spec-Driven Development workflows, curated coding skills, MCP servers, and per-phase model assignment. It supports 15 AI agents and includes delegation triggers to keep sessions organized—such as delegating when reading 4+ files, touching 2+ non-trivial files, or after commits and PRs. Key features include OpenCode SDD Profiles for assigning different models to different development phases, persistent memory that remembers decisions across sessions, and a skill registry system. The project provides installation via Homebrew or Scoop, workspace-scoped or global configuration options, automatic config backups with pruning, and an interactive TUI for management. Documentation covers intended usage patterns, SDD profile management, Engram commands for memory browsing, and agent-specific setup guides.

The website content is in Hindi and discusses "vibe coding" - creating applications with AI assistance. The video covers creating a game application using AI Studio (Google's tool) with an infinite scroll mechanic designed to be addictive. The creator emphasizes the importance of first developing a detailed PRD (Product Requirements Document) using ChatGPT before implementation, then breaking down the full specification into sequential prompts for Claude to execute step-by-step. The approach highlights using discussion mode in Claude when encountering errors rather than direct implementation attempts. For monetization, the creator integrates AdSense ads into the Android app. The workflow demonstrates practical agentic coding by using AI to handle the development iteratively while the human maintains strategic control through well-structured prompts. The creator also mentions using a Notion document to organize and share all prompts used in the process, emphasizing that new applications can be quickly recreated and iterated based on market feedback.

Based on the user's interests in agentic coding, AI agents with Claude/Cursor, PRD workflows, and advanced vibe coding techniques, here is the relevant news summary: Agnes AI founder Bruce Yang discusses how AI coding agents are evolving beyond simple code completion into autonomous development. Engineers at Agnes AI are now using tools like Claude Code and Cursor with a philosophy of 10% human alignment time followed by agents autonomously building entire codebases over several hours, effectively creating scenarios where one person directs multiple AI agents working in parallel. Yang emphasizes that vibe coding is transitioning from intermediate autocomplete features toward true autonomous agent workflows, and notes that coding represents one of the major AI directions for the year with potential to achieve AGI-level capabilities. He highlights how agentic workflows enable cost-effective application development at roughly 1/100th to 1/200th the cost of competitors like OpenAI and Anthropic, making it feasible for solo entrepreneurs and small teams to build multiple products and run AB tests without token usage concerns. Agnes AI is developing its own coding harness launching next month to integrate with existing tools like Claude Code and Cursor through open APIs.