{
  "markdown": "# Skills\n\nMy personal collection of AI agent skills.\n\n## Overview\n\nThis repository contains reusable skills that extend AI agent capabilities. Each skill is a self-contained module that teaches an agent how to perform specific tasks.\n\n## Project Structure\n\n```\nskills/\n├── arxiv-paper-report/  # arXiv field-intelligence and decision report skill\n├── blueprinter/          # Technical diagram generation skill\n├── brief/                # Brief site template and report publishing skill\n├── design-hub/           # Local design-system library and extraction skill\n├── ten-defeats/           # Judgment and completion guardrail skill\n└── README.md\n```\n\n## Available Skills\n\n| Skill | Description |\n|-------|-------------|\n| [arxiv-paper-report](./arxiv-paper-report/) | Turn an arXiv-centered literature corpus into evidence-bounded field beliefs, capability frontiers, technical options, and explicit decisions in validated JSON and standalone HTML |\n| [blueprinter](./blueprinter/) | Generate technical diagrams in Flat Engineering Blueprint style using HTML/CSS |\n| [brief](./brief/) | Initialize a Brief React/Vite report site, generate analysis reports, and publish them into the site |\n| [design-hub](./design-hub/) | Initialize and manage a local Design System Hub, extract systems from screenshots, and expose Agent-readable exports |\n| [ten-defeats](./ten-defeats/) | Guard non-trivial software work against action theater, tool-substituted design, proxy validation, and premature completion |\n\n### arXiv Decision Intelligence Report\n\nUse `arxiv-paper-report` when the goal is not merely to collect or summarize papers, but to decide what a vertical field currently supports. It turns atomic source evidence into testable propositions, demonstrated capability boundaries, a visibly labeled mechanism model, comparable or non-comparable technical options, transition signals, five-dimensional maturity, decision consequences, and leading indicators.\n\nThe report is deliberately explicit when the evidence cannot support an authoritative view, historical trend, deployment claim, or route ranking. Its primary artifacts are a strict JSON research record and a deterministic, bilingual standalone HTML report.\n\n![Direct answer and epistemic ceiling](./images/arxiv-paper-report-direct-answer.png)\n\n![Capability frontier with source evidence](./images/arxiv-paper-report-capability-frontier.png)\n\n![Source and analyst mechanism model](./images/arxiv-paper-report-mechanism-model.png)\n\nExample requests:\n\n- \"Research proactive AI agents and tell me what is demonstrated, what remains unknown, and what we should validate next.\"\n- \"Map the technical frontier of this field and produce an evidence-traceable Chinese/English HTML report.\"\n- \"Compare the available technical options without ranking approaches that were evaluated on incompatible tasks.\"\n- \"Determine whether recent papers represent a structural transition or only a cluster of submission signals.\"\n\nSee the [skill README](./arxiv-paper-report/README.md) for its evidence model, dependencies, artifact contract, and local validation commands.\n\n### Example: Blueprinter Output\n\n![Claude Code Architecture Blueprint](./images/claudecodesource.png)\n\n### Brief\n\nUse the `brief` skill to create and operate a professional research brief site. It can initialize a new Brief site, turn source material into standalone Brief-style analysis reports, publish those reports into the site, and start the local preview server.\n\n[Live example](https://brief.blueprinter.ai/)\n\n![Brief site preview](./images/brief.png)\n\nExample requests:\n\n- \"Use the brief skill to initialize a new site and start it locally.\"\n- \"Use brief to analyze this article, generate a report, and publish it to my Brief site.\"\n- \"Use brief to add this codebase analysis as a new report and make it the featured report.\"\n- \"Use brief to build the site and tell me where the production output is.\"\n\n### Design System Hub\n\nUse the `design-hub` skill to initialize and operate a local React/Vite design-system library. It can extract a coherent system from screenshots or written direction, generate a source-specific and privacy-safe Live Mock, and expose Copy Tokens, `DESIGN.md`, standalone HTML, and an Agent-oriented API document.\n\n![Design hub preview](./images/designhub.png)\n\nNew sites default to `~/design-hub` and include the current MuleRun, Raft, QoderWork, and Superset design systems. New systems are appended as isolated entries without rewriting existing content.\n\nExample requests:\n\n- \"Use design-hub to initialize the local site and start the service.\"\n- \"Extract a design system from these screenshots and add it to my Hub.\"\n- \"Regenerate this system's Live Mock while preserving the existing entries.\"\n- \"Export the design system as DESIGN.md and show me the Agent API endpoint.\"\n\n## License\n\nMIT\n## Star History\n\n[![Star History Chart](https://api.star-history.com/image?repos=yofine/skills&type=date&legend=top-left)](https://www.star-history.com/?repos=yofine%2Fskills&type=date&legend=top-left)\n\n<br/>\n",
  "bytes": 5045,
  "sha": "93df3c8707f52cb8362b74bce8d5e2aec7782cfb3f46fc00233f3873534d37c0",
  "repo_slug": "yofine/skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_yofine_skills_blueprinter_a7461740/readme"
}