BuilderIO/skills

Skills for agents

What it solves

This project provides a library of small, composable "skills" designed to extend the capabilities of AI agents (such as Claude Code, Codex, and others). It addresses common agent limitations, such as the tendency to guess based on stale memory instead of reading current documentation, the difficulty of visualizing complex plans or code diffs, and the inefficiency of using expensive frontier models for routine, token-heavy tasks.

How it works

Skills are implemented as modular instructions and tools that can be installed via a CLI or as plugins for specific agent environments. They enable agents to perform specialized actions, such as:

  • Software Lifecycle Management: The /factory suite automates the pipeline from collecting telemetry and feedback to reviewing and shipping PRs.
  • Visual Interfaces: Tools like /visual-plan and /visual-recap convert text-based plans and diffs into interactive MDX-based visual summaries.
  • Agent Orchestration: Skills like /efficient-fable and /efficient-frontier allow a high-cost model to act as an architect while cheaper agents handle the heavy lifting.
  • Environment Integration: Skills like /an and /webmcp allow agents to operate Agent-Native apps and web apps through MCP (Model Context Protocol) tools.
  • Context Recovery: The /rewind skill integrates with local screen memory to recover recent visual and textual context.

Who it’s for

Developers and teams using AI agents for coding and software delivery who want to increase agent autonomy, reduce token costs, and improve the human-in-the-loop review process through visual aids.

Highlights

  • Factory Workflow: An experimental system for turning telemetry and errors into policy-gated software changes.
  • Visual Planning & Recaps: Transforms abstract text plans and raw git diffs into scannable, interactive visual maps.
  • Cost Optimization: Frameworks for preserving expensive models for judgment while using cheaper models for execution.
  • MCP Integration: Deep integration with the Model Context Protocol for operating web and native apps.
  • Autonomy Tools: Skills like /plow-ahead and /agent-watchdog for handling ambiguity and auditing other agents.

Related

  • Project
    chujianyun/skills

    A Chinese‑language repo that bundles dozens of reusable “Agent Skills” – small, documented plugins for Claude Code and similar LLM‑agent platforms. Skills cover wiki look‑ups, prompt optimisation, career assessments, diagram generation, document interpretation, audio‑to‑text, image watermarking, web‑scraping, and service run‑books. They are organized under `skills/`, registered in `config/skills.json`, and installed via Claude Code’s marketplace, an installer agent, or the `openskills` CLI.

  • Project
    kangarooking/kangarooking-skills

    A collection of structured, open-source Agent Skills compatible with the Agent Skills standard, providing reusable workflows for content creation, image generation, and AI project engineering.

  • Project
    Pluviobyte/rnskill

    A comprehensive library of modular AI Agent skills for Codex and Claude Code, specializing in automated content creation pipelines for video, social media, and business analysis.

  • Project
    badlogic/pi-skills

    A collection of extensible skills for AI coding agents that enable capabilities like web search, browser automation, and Google Workspace integration.

  • Project
    DannyMac180/skills

    A collection of reusable AI agent skills and plugins for Claude Code and Codex, enabling advanced supervised workflows, mod management, and personalized learning.