搜索

Evidence-first agent skills for handoff auditing and documentation sync, packaged for Codex and DeepSeek Harness.

AI Agent Skills 合集 — CMD+PowerShell 避坑指南 · CLI 优先决策矩阵 · 更多技能持续添加

专门适配于 DeepSeek v4 pro 正式版 与 DeepSeek Harness 上的小说创作插件：让 AI 写小说，但读者闻不出 AI 味。

Portable Craft-style Mermaid generation, rendering, and visual review skill for AI coding agents

A tool box for literature search and analysis

End-user presentation workflows, preview application, and Agent Skill powered by PPTKit.

AI 足迹 MCP Server — 让 AI Agent 管理用户的数字足迹

marketplace

面向 AI 编程代理的 Android 功能复刻 Skill。它以 Android 源项目的实际实现为依据，将指定功能完整迁移到其他项目或平台，并保持功能链路、业务行为、UI 和可复用资源一致 An agent skill for reproducing an existing Android feature in another project or platform. It treats the Android implementation as the sour

JARVIS is an intelligent layer that gives agents access to their code, knowledge, context, memory, tools, and runtime — starting locally on your machine, with the ability to extend into the cloud.

bloome-finance-plugin

High-performance, capability-aware repository tools and process execution for coding agents.

Bridge vision to text-only coding agents — a Claude Code & Codex plugin for DeepSeek and friends

Persistent memory layer for AI agent swarms. Postgres-backed retrieval, Markdown as source of truth, sub-second cascade sync. Self-hosted.

Design-aware motion engineering for coding agents. 动效，是时间的排版。26-scene dispatch, engine routing, quality gates, independent review.

An LLM-driven Minecraft bot that beats the Ender Dragon. Rust + Azalea protocol client, 44 tools, 23-tier task system.

workled skill

scip-swift

openin

Vision-first local MCP that gives text-only Agents image understanding through Agnes AI (BYOK).

DeepSeek Harness (dsh) 插件：把 LLM agent loop 中确定性的执行路径编译成 DSL 程序并直接执行，显著降低 token、往返轮次与上下文暴露。A DeepSeek Harness plugin that compiles deterministic agent paths into DSL programs.

Trust checkpoint for AI-generated code, agent reports & embodied-AI artifacts — the exam hall for LLM output. Solves LLM hallucination with programmatic verification: four-layer statistical exams, invariant examiner, conclusion anchors. No

Tkinter/ttk 原生桌面应用全生命周期技能（MVC + pygubu .ui + SQLite + PyInstaller）

FastHTML + pywebview 桌面应用全生命周期技能（本地 HTTP + WebView2 + PyInstaller）