Pentest-Swarm-AI is an open-source autonomous penetration testing tool built on a stigmergic swarm architecture. It coordinates multiple LLM-driven specialist agents (recon, classifier, exploit, report) via a shared blackboard with pheromone decay, enabling emergent attack paths without a central planner. The system integrates 30+ external security tools (nmap, nuclei, sqlmap, etc.), multiple LLM providers (Claude, OpenAI, Ollama, Gemini, LM Studio), and bug bounty platforms (HackerOne, Bugcrowd, Intigriti) for automated scope import and report submission. The architecture is a multi-agent system with a scheduler that dispatches agents based on blackboard predicates and pheromone levels. Key entry points include a CLI (Cobra commands: scan, init, playbook, submit, etc.), an API server (Fiber), a WebSocket hub for live streaming, a Next.js web dashboard, and an MCP server for IDE integration. Data flows: agents write findings to the blackboard with pheromone weights; the scheduler reads the blackboard and triggers agents whose predicates match; agents read findings, process them, and write new findings, creating a stigmergic feedback loop. External dependencies include PostgreSQL with pgvector for durable blackboard storage, LLM API endpoints, and the installed security tool binaries. The codebase demonstrates high craftsmanship with clean abstractions (tool adapter pattern, LLM provider interface, blackboard interface), extensive testing (unit, integration, e2e, benchmarks, LLM eval), and safety features (scope enforcement, provenance signing, rate limiting, false-positive caching). It is actively maintained with recent commits, CI/CD pipelines, and community engagement.
TraceAI is a multi-language (Python, TypeScript, Java, C#) OpenTelemetry-based instrumentation SDK for AI/LLM applications. It provides drop-in auto-instrumentation for 50+ LLM providers, agent frameworks (LangChain, LlamaIndex, CrewAI, AutoGen, etc.), and vector databases (Pinecone, ChromaDB, Qdrant, etc.). The core library wraps SDK calls to capture structured spans with semantic attributes (tokens, model, messages, tool calls) and exports them via OTLP to any OpenTelemetry-compatible backend, with a default hosted backend at Future AGI. The architecture is a monorepo with per-framework packages that follow a consistent instrumentor pattern: register a tracer provider, instrument the target library, then make API calls. Key entry points are `register()` (Python/TypeScript), `TraceAI.init()` (Java), and `TraceAI.Register()` (C#). The SDK also includes context attribute propagation, PII redaction, evaluation tags, and configurable masking. While the codebase is well-structured and actively maintained, there is significant duplication across framework instrumentors and some tests rely on heavy mocking, which limits reusability of individual components.
Exosphere is a reliability runtime for AI agents, providing a state-based workflow orchestration platform. The system consists of three main components: a Python SDK (`exospherehost`) for defining nodes and execution graphs, a FastAPI state-manager backend that handles state transitions, retries, cron triggers, and secret encryption, and a Next.js dashboard for monitoring and managing workflows. The architecture follows a microservices-like pattern within a monorepo, with the SDK acting as a client library and the state-manager as the central orchestration engine. Key capabilities include dynamic graph execution, infinite parallel fanout/unite patterns, configurable retry policies (exponential, linear, fixed with jitter), signal-based state pruning and re-queuing, and a graph-level key-value store. The system relies on MongoDB for persistence and uses Beanie ODM. The codebase demonstrates solid engineering practices with well-structured controllers, middleware for request ID propagation and exception handling, and a comprehensive test suite, though some tests are incomplete or skipped. The project is in early beta (v0.0.3b1) with active development and monthly releases.
Bluebox is a library and agent framework for reverse engineering web applications by monitoring browser interactions via Chrome DevTools Protocol (CDP). It captures network requests, DOM snapshots, storage changes, and user interactions into JSONL files, then uses LLM-driven agents to discover reusable automation routines (parameterized workflows) that can navigate, fetch, click, type, and extract structured data from live websites. The system includes a conversational agent (BlueBoxAgent) that searches a routine index and falls back to an AI browser agent, a GuideAgent for interactive routine creation, and a PrincipalInvestigator orchestrator that runs parallel specialist agents (network, DOM, storage, UI) to build routine catalogs from captured sessions. Architecturally, Bluebox is a modular Python application with clear separation: CDP monitors and session management, Pydantic data models for routines and events, abstract base classes for agents and data loaders, LLM vendor clients (OpenAI, Anthropic), a workspace abstraction for artifact management, and CLI/HTTP entry points. The SDK provides high-level methods (monitor_browser, discover_routine, execute_routine) and the package ships with example routines for real-world sites (Amtrak, Polymarket, Massachusetts corporate search). While primarily an application/library, Bluebox has a strong data pipeline character: it ingests raw CDP events, transforms them through data loaders and vectorstores, and produces structured routine definitions. The benchmark framework evaluates routine quality with deterministic and LLM-based tests. The codebase demonstrates high craftsmanship with reusable patterns (agent tool decorator, workspace abstraction, sandboxed execution, data loader interface) and extensive unit tests, though some specialist agents are marked as beta and the orchestration pipeline is complex.
Agentic Awesome Skills is a curated marketplace infrastructure for AI coding agent skills—instructional playbooks (SKILL.md files) with accompanying scripts and assets. The repository provides a multi-platform distribution system supporting Claude Code, Cursor, Gemini CLI, Codex CLI, Kiro, and others. It includes a React+Vite web application for browsing, searching, and installing skills; a CLI and MCP server for local agent integration; and a comprehensive validation, security scanning, and release pipeline. The codebase implements a transaction engine with deterministic content addressing (canonical JSON + SHA-256), durable state management, and schema-validated skill installation, forming a robust foundation for agent-owned tool selection and workflow planning. While the core infrastructure (AAS v1 library, MCP server, catalog search) demonstrates high craftsmanship and reusability, the repository is hampered by massive file duplication: identical skill scripts and governance modules appear under `skills/`, `plugins/agentic-awesome-skills/`, and `plugins/agentic-awesome-skills-claude/`, creating synchronization risk. Some skills contain hardcoded paths or incomplete boilerplate. The repository is actively maintained with CI/CD, automated audits, and security scrutiny, though the sheer volume of 2,474+ skills imposes ongoing maintenance burden. Overall, it is a well-engineered, production-quality skill distribution platform with notable reusable design patterns.
Vibe Kanban is a full-stack, dual-deployment application (local Tauri desktop app + remote cloud service) that combines Kanban-style issue tracking with AI coding agent workspaces. The local backend is a Rust Axum server backed by SQLite, while the remote backend uses PostgreSQL with ElectricSQL for real-time sync. The system allows users to create projects, manage issues, and launch coding workspaces where AI agents (Claude, Codex, Gemini, Copilot, etc.) execute tasks on isolated branches with preview servers, git operations, and automated code review. It includes a relay tunnel system for remote host pairing, a WebRTC transport for low-latency streaming, and an MCP server for external tool integration. The frontend is built with React, Tailwind CSS, and a shared component library, supporting localization in 7 languages. The project is officially sunsetting, as stated in the README, but the codebase remains a substantial example of a modern Rust/TypeScript monorepo with real-time collaboration, AI executor abstraction, and multi-platform delivery.
This repository is a TypeScript monorepo for an AI-powered autonomous web browsing agent called 'open-browser'. It consists of three packages: core library (open-browser), CLI (@open-browser/cli), and sandbox (@open-browser/sandbox). The core library orchestrates an agent loop that uses Playwright to control a Chromium browser, extracts DOM state via CDP, sends it to an LLM (OpenAI, Anthropic, Google) which decides actions in structured JSON, and executes those actions via a CommandExecutor. The system supports vision screenshots, extended thinking, compact mode, planning, evaluation, replay recording, and MCP server integration. The CLI provides commands for running agents, interactive sessions, and direct browser manipulation. The sandbox provides resource-limited execution. Key entry points include the Agent class, Viewport browser session, CommandExecutor, and BridgeServer. External dependencies include Playwright, Vercel AI SDK, Zod, Turndown, and various LLM provider SDKs. The architecture follows a modular monorepo with clear separation of concerns, event-driven communication via EventHub, and a guard/watchdog pattern for browser lifecycle management. The codebase demonstrates strong engineering practices with extensive use of Zod schemas for structured outputs, a command catalog pattern, and comprehensive telemetry and cost tracking.
This repository is a static website for a Chinese-language educational course titled 'Deep Agents 实战' (Deep Agents in Action). It provides 16 chapters of technical content on building AI agents using LangChain and LangGraph, including text, video links, and downloadable PDFs. The site is built with Astro 6, Tailwind CSS 4, and TypeScript, and is deployed to GitHub Pages via GitHub Actions. The codebase is well-structured: content is authored in Markdown under `content/`, metadata is managed in `scripts/chapters.json`, and a preprocessing script injects frontmatter into the Astro content collection. Custom Astro components handle layout, navigation, theme toggling, and a responsive table of contents. The repository also includes CI/CD workflows for asset validation and cross-platform installation testing. While the primary purpose is educational content delivery, the build and content pipelines demonstrate reusable patterns for static site generation with Astro.
This repository provides a collection of 52 specialized AI agent definitions for penetration testing and red team operations, designed to be used with Claude Code. It is distributed as a Claude Code plugin, with an interactive installer (install.sh), a SQLite findings database (db/findings.sh) for persisting engagement data, and a versioned migration system. The agents are defined as Markdown files with YAML frontmatter, each containing structured system prompts that enforce scope validation, OPSEC tagging, evidence handling, and MITRE ATT&CK mapping. The system includes a swarm orchestrator for multi-agent coordination, a scope guard for safety, and a two-tier safety model (advisory vs. Bash-capable agents). While primarily a configuration/instruction set rather than a traditional software application, it demonstrates strong engineering practices in prompt architecture, database schema design, and distribution automation. The repository is actively maintained with CI validation, shellcheck, and unit tests for the database scripts.
This repository is the full-stack implementation of Mega, a large-monorepo Git hosting and code collaboration platform with a Rust backend and a large React/Next.js frontend. The backend is split into several crates: mono provides the HTTP/SSH API server, ceres contains application services for Git operations, change lists (CLs), code review, build triggers, webhooks and notifications, jupiter handles persistence through SeaORM entities and storage services, jupiter-migrate manages a large sequence of database migrations, and orion/orion-server implement a Buck2-based distributed build orchestration system. The frontend in moon/apps/web is a feature-rich collaboration application with posts, notes, chat, notifications, project management, diff viewing, CL/merge-queue review, admin groups, OAuth integrations, and editor support. The core operational flow is: git push/import -> ceres transport and jupiter storage -> change list and merge-queue workflows -> build triggers sent to Orion -> build events/checks streamed back to the UI.
This repository provides a Python library for converting web pages into text optimized for Large Language Model (LLM) input. It fetches HTML content using Playwright (headless browser) and processes it to extract meaningful text while preserving structural information like tables, links, and images in a text format. The core processing logic converts HTML tables to markdown, handles image and link preservation (or removal), and strips unnecessary tags (script, style) to reduce noise. The library is designed as a preprocessing step before feeding web content to LLMs for tasks such as data extraction or summarization. It includes three main modules: HTML fetching, HTML processing, and a combined convenience function.
Spawn is a multi-cloud VM provisioning CLI that deploys AI coding agents (Claude Code, Codex, OpenClaw, etc.) onto cloud VMs (Hetzner, DigitalOcean, AWS Lightsail, GCP, Daytona, Sprite) with a single command. It handles the full lifecycle: interactive agent/cloud selection via fuzzy matching, VM provisioning, SSH key injection, agent installation (npm binaries, pre-built tarballs, or Docker sidecars), environment configuration (OpenRouter API key, MCP servers via skill files), and interactive session handoff. The system supports headless mode, recursive spawning (agent launching sub-agents), auto-update systemd services, and a security scan cron. It is distributed as per-cloud JS bundles (Bun.build, minified) with curl|bash wrappers that download from GitHub Releases, and a Windows PowerShell installer. Architecturally, Spawn is a modular monolith: a shared orchestration pipeline (`shared/orchestrate.ts`) drives cloud-specific providers implementing a common `CloudRunner`/`CloudOrchestrator` interface (authenticate → promptSize → createServer → waitForReady → install → configure → launch → interactiveSession). Agents and clouds are defined in a `manifest.json` that validates compatibility via runtime type contracts. A shared `Result` monad (`@openrouter/spawn-shared`) enables functional error handling with retry semantics across SSH calls and cloud API interactions. The codebase is well-tested (~100+ test files) with fs-isolated sandboxes, mocked cloud prompts, and integration tests against real cloud providers via an E2E harness. The codebase demonstrates strong engineering practices: schema-validated input parsing (valibot), security-first validation (allowlists for every user-controlled input: instance names, IPs, SSH usernames, shell commands, remote paths), and atomic file operations with corruption recovery for history persistence. However, there is significant duplication across cloud providers (near-identical SSH execution patterns in each cloud module), large monolithic files (orchestrate.ts ~1000+ lines, agent-setup.ts ~2000 lines), and manual drift risk from hardcoded API slugs (DigitalOcean marketplace images, Hetzner server types, GCP machine types) and embedded OAuth client secrets (documented as acceptable public-client pattern).
This repository is a comprehensive collection of 21 self-contained AI agent implementations, each solving specific business or technical problems using various frameworks (LangGraph, CrewAI, AutoGen, Agno, LlamaIndex). Each agent is designed to be runnable end-to-end in under 10 minutes with its own dependencies, configuration, and documentation. The repository also includes a React-based web application that serves as an atlas for browsing and learning about these agents, plus an educational course on CrewAI with MCP integration. The architecture consists of independent CLI applications (the agents) coupled with a documentation/web layer that reads agent metadata for display purposes.
The Prompt Engineering Guide is a comprehensive, multilingual documentation site built with Next.js and Nextra that serves as an educational resource for prompt engineering with large language models. While the site itself is a web application, its core value lies in the extensive collection of tutorials, technique explanations, model-specific guides, and research summaries that constitute a structured knowledge base on effective LLM prompting. The repository includes interactive Jupyter notebooks that demonstrate practical applications of techniques like PAL, RAG, and ReAct, alongside hundreds of MDX pages covering topics from basic prompting to advanced agent building. Content is available in numerous languages, reflecting a commitment to global accessibility. The site features custom UI components for enhanced usability, such as announcement bars, course promotions, and copy-to-clipboard functionality, but these serve primarily to facilitate access to the educational content. Despite being actively maintained, the guide faces challenges typical of rapidly evolving fields: content requires frequent updates to keep pace with new LLM releases and research findings, and the multilingual nature increases the risk of translation drift. External links to papers, videos, and tools may also become stale over time. Nevertheless, the repository remains a valuable reference for practitioners and researchers seeking to understand and apply prompt engineering principles.
This repository is an 'awesome list' — a static Markdown index of externally hosted Claude Code skills, organized into 14 categories (documents, development, data, science, writing, learning, media, health, collaboration, security, utility, articles, and collections). It contains no source code, no executable skills, no CI/CD, and no tests. The production function is discovery: a human or AI agent reads the README to find third-party skill repositories, then clones or installs those skills separately. Architecture is a single README with external links plus a few embedded sponsor images. There are no data flows or APIs. The list is actively maintained with a recent commit and a large volume of open issues, indicating ongoing community input. The core value is the curation and categorization of 150+ external skills, not software engineering craftsmanship. Sponsored placements appear at the top, so it is not purely community-driven. Overall, a useful and well-organized directory, but with zero in-repo technical content.
agentsre is a Python library that applies Site Reliability Engineering (SRE) practices to AI agent systems. It provides reliability instrumentation for production agents: semantic validation of agent outputs, SLI metrics collection, circuit breaking, escalation baseline detection, cost tracking, and structured audit logging. The codebase is organized as a package with a core reliability module (aos_reliability.py), an orchestration subpackage for control-plane SLIs, a validation layer for agent decisions, and a cost optimizer for LLM spend. It also includes a fintech orchestration demo and documentation for running agents in payment systems. Architecturally, it is a library rather than a service: consumers import the Python API and embed checks into their agent loop. The primary data flow is agent action -> semantic validation -> reliability/risk checks -> orchestration SLI recording -> audit log. The system tracks metrics like escalation rates and decision quality, and can block or allow requests via a circuit-breaker style interface. External dependencies are minimal, with optional cost-optimizer imports guarded by try/except. Maintenance is active, with CI workflows, tests, and a changelog. The code is well-documented and the modules are mostly self-contained, making several components extractable for reuse in other agent platforms.
This repository is a curated collection of Claude skills—markdown instructions and supporting scripts that enable Claude (or any MCP-compatible agent) to perform a wide range of tasks, from document manipulation (DOCX, PPTX, PDF, XLSX) and GIF animation to web testing, MCP server building, and automation of hundreds of third-party SaaS services via Composio's Rube MCP. The repo is primarily a skill library with a standardized template for each skill, but it also contains several well-crafted Python libraries (e.g., easing functions, particle systems, OOXML validation frameworks) that are genuinely reusable. The architecture is modular: each skill lives in its own directory with a SKILL.md and optional scripts. The heavy reliance on external services (Composio, Rube MCP) and the massive duplication of templated skill documentation introduce maintenance risks, but the core utility scripts demonstrate solid engineering craftsmanship.
This repository is a reference architecture and plugin collection for deploying AI agents in financial services using Anthropic's Claude platform. It defines 10+ named agents (e.g., Pitch Agent, Earnings Reviewer, KYC Screener) and 100+ skills that guide Claude to produce institutional-grade financial documents: Excel valuation models (DCF, LBO, comps), PowerPoint pitch decks, Word research reports, and HTML earnings previews. The architecture follows a dual-delivery model: agents can be installed as Claude Cowork plugins via marketplace registration, or deployed headlessly via the Managed Agents API (agent.yaml manifests in managed-agent-cookbooks). Skills are authored in vertical plugins organized by domain (investment banking, equity research, private equity, wealth management) and synced into agent bundles. The repository also includes a Microsoft 365 add-in installer (claude-for-msft-365-install) for deploying Claude into Excel, Word, PowerPoint, and Outlook. Data flows rely heavily on MCP servers (LSEG, S&P Global, Daloopa, FactSet) and SEC EDGAR for financial data, with strict citation and formula-over-hardcode conventions. The repo is primarily prompt engineering and configuration rather than application runtime code, but it includes several reusable Python scripts for skill scaffolding, document validation, and cross-reference linting.
SenseNova-Skills is a composable collection of AI agent skills for office automation, built around the SenseNova model family. It provides production-grade capabilities for image generation and recognition, presentation (PPT) creation, Excel data analysis, deep research with structured evidence, and multi-source search (academic, social, code, finance). The skills are designed to run in Agent Skills-compatible runtimes (OpenClaw, hermes-agent) and are bundled into the Raccoon product for enterprise use. The architecture is modular: each skill is a self-contained directory with a SKILL.md manifest, scripts, prompts, and schemas. The deep research skill implements a multi-agent pipeline with formal JSON schemas and validators. The PPT skill has a multi-stage pipeline from user input to HTML-to-PPTX conversion via Playwright. The codebase demonstrates solid engineering with reusable components like a shared LLM/VLM client, export pipeline libraries, and validation infrastructure. However, it suffers from duplicated utility code across skills, complex environment variable fallback chains, hardcoded paths, and dependency on external APIs and headless browsers. The repo is actively maintained with biweekly reports and frequent commits.
This repository is a monorepo of self-contained demo applications showing how to build and deploy multi-agent AI systems using Docker Compose, Docker Model Runner, and the MCP Gateway. It contains 12+ independent examples spanning Google ADK, A2A, Agno, CrewAI, LangGraph, Langchaingo (Go), Spring AI, Akka, and a Next.js agent UI. Each subproject is a standalone composition with its own compose.yaml, Dockerfile, and README, designed to run with a single `docker compose up --build` after creating a `.mcp.env` file. The demos are reference architectures rather than a single product, emphasizing local LLM inference, MCP tool integration, and switching between local and cloud models via compose override files.