262 boosters for "ai-agents" — open source, verified from GitHub, ready to install
"name": "director-mode-lite", "description": "Use Claude Code like a Director, not a Programmer. Complete toolkit with 26 commands, 14 agents, 31 skills, and TDD-based Auto-Loop.", "name": "Lucas Wang",
A unified initialization booster that streamlines project setup by routing to specialized commands for documentation, preferences, and design tokens—ideal for developers standardizing new projects and maintaining consistent configurations.
"description": "You say it, AutoCode ships it. 47 skills. Code to deployment in one session. I-Lang v3.0 powered.", "name": "I-Lang Research, Eastsoft Inc.", "url": "https://github.com/ilang-ai/autocode"
"name": "use-agently", "description": "Use Agently CLI", "homepage": "https://use-agently.com",
"description": "Ship software systematically: project lifecycle, TDD, parallel agents, code review, security auditing, and infrastructure validation", "email": "lgbarn@users.noreply.github.com"
This MCP server integrates Nutrient's DWS Processor API with Claude, enabling AI agents to perform advanced document processing tasks including PDF handling, OCR, digital signatures, and redaction. Developers building document automation workflows and AI-powered applications benefit from seamless integration with Claude's context protocol.
A comprehensive tutorial chapter teaching developers how to build and deploy intelligent AI agents with tool use and automation capabilities in AnythingLLM. Ideal for engineers looking to add autonomous workflows and function-calling to their self-hosted RAG systems.
Parse PDF, Office (Word/PPT/Excel), and image files into clean Markdown — with LaTeX formulas, tables, images, and OCR. One zero-dependency script, two backends, No , no API key. The free Agent API handles files ≤ 10 MB / ≤ 20 pages.
"name": "ai-marketing-skills", "name": "SuperAmped", "url": "https://superamped.com"
This project includes deepsafe-scan, a preflight security scanner for AI agent environments. When the user asks to audit, scan, or check security of their AI agent setup, skills, or MCP servers, run the scanner: For a full scan with LLM analysis (if OPENAIAPIKEY is set):
This skill describes the current runtime line. It intentionally does not describe the old Python , completion-file finalizer, or JSON delivery-outbox runtime. Those were historical mechanisms and must not be
"name": "alfred-dev", "description": "Plugin de ingeniería de software completa: 10 agentes de núcleo y 8 opcionales con personalidad propia, memoria persistente por proyecto, quality gates y flujos automatizados desde la idea hasta producción.", "homepage": "https://alfred-dev.com",
An MCP server that integrates Crawlbase API for web scraping capabilities into AI agents and code editors like Claude, Cursor, and Windsurf. Developers building AI applications needing reliable web data extraction benefit from this seamless integration.
"name": "@kastalien-research/thoughtbox", "packageManager": "pnpm@9.15.4", "description": "MCP server providing cognitive enhancement tools for LLM agents: structured reasoning, mental models, and literate programming notebooks",
人脑不是硬盘，不会无差别存储所有信息。人脑会： DNA Memory 模拟这个过程，让 Agent 真正"进化"。 并且仅在有新的 写入后才执行 ，避免重复归纳同一批记忆。
If no interval is specified, the default interval is 10 minutes. The loop stops automatically when any of the following occur: 1. Maximum iteration count (100) is reached
Two tiers: hooks handle automatic context flow (surfacing, extraction, compaction survival). MCP tools handle explicit recall, write, and lifecycle operations. Three instances for neural inference. The wrapper defaults to . All three models auto-download via if no server is running (Metal on Appl
"name": "espalier-engineering", "description": "Train your AI coders the way you'd train a vine — discover your codebase's actual patterns, then encode them as rules, skills, agents, hooks, and a guided pipeline so generated code lands inside your conventions on the first try, not the fifth", "homep
"description": "2389.ai skill collection including Firebase, CSS, testing, and development workflows",
"name": "wanderlog-mcp", "description": "MCP server for Wanderlog — let Claude build and edit your trip itineraries through conversation", "wanderlog-mcp": "dist/index.js"
"name": "dcode-agent-kit", "description": "Scaffold a Deep Agents (LangChain) agent or a dcode CLI agent from one command (/new-dcode-agent). The agents it builds work with any OpenAI-compatible provider.", "author": { "name": "Elia Alberti" },
"name": "rk-skills", "displayName": "rk-skills", "version": "1.17.1",
"$schema": "https://json.schemastore.org/claude-code-plugin-manifest.json", "description": "Animated data-flow diagrams your AI agent can write. Written by AI. Watched by humans.", "name": "Naor Sabag",
1. 架构不是「画」出来的,是从约束里「逼」出来的。 没搞清约束就画图,画什么都是瞎画。 2. 没有银弹,只有取舍。 任何决策本质都是「用 A 换 B」。一个「没有缺点」的方案,不是完美,是没想清楚。 3. 没有「最好的架构」,只有「在这组约束下最合适的架构」。 同样是聊天,内部工具和微信的答案天差地别。
