planning-with-files is a persistent, file-based task-planning system for AI coding agents. It maintains task_plan.md, findings.md, and progress.md on disk and wires into agent lifecycle hooks (SessionStart, UserPromptSubmit, PreToolUse, PostToolUse, PreCompact, Stop) across Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, Pi, Hermes, and a dozen other assistants, re-injecting plan context into the model's context window on every turn. Plans survive context loss, /clear, compaction, and crashes; a session-catchup mechanism recovers prior transcripts. A v3 gated mode can block the agent from stopping until plan phases complete, backed by an append-only JSONL run ledger, stall detection, and SHA-256 plan attestation that makes hooks refuse to inject tampered content. Architecturally it is a distribution hub rather than a single application: a canonical skill under skills/planning-with-files is mirrored byte-identical into 15+ IDE-specific directories, with native TypeScript plugins (OpenCode, Pi, DSH), a Python plugin (Hermes), and shell/PowerShell hook dispatchers. Plan resolution is a fail-closed chain (PLAN_ID env β .active_plan pointer β newest plan dir β legacy root) guarded by slug validation, containment checks, symlink/junction rejection, and strict PWF_PLAN_ROOT pinning. The engineering is unusually rigorous for an agent-skill repo: 400+ pytest tests, byte-parity tests between shell/PowerShell/Python/TypeScript twins, a PATH-stub BSD userland simulation to catch GNU-only flags that silently fail on macOS, and heavy Windows PowerShell 5.1 / Git Bash portability hardening. Non-obvious patterns include KV-cache-stable injection output (no timestamps or free text in summaries, wall-clock normalization to T00:00:00Z) and an mtime-keyed attestation cache for performance. The repository is a developer tool, not a data/research project; although it ships benchmark and eval documentation (assertion pass rates, recovery-turn comparisons), those are verification evidence for the skill, not analysis artifacts. It is actively maintained at v3.24.0 with recent commits, 4 open issues, and 100+ releases; technical drift is low, with the main maintainability risk being deliberate duplication across 20+ canonical mirrors, mitigated by sync-ide-folders.py and parity tests but still a possible source of silent drift.
Oh-My-ClaudeCode (OMC) is a multi-agent orchestration layer for Claude Code, Anthropic's AI coding assistant. It extends Claude Code with a plugin system, custom MCP tools, a team collaboration engine, a real-time HUD, and a rich hook system. The system is implemented in TypeScript and runs as a CLI tool (`omc`), an npm package (`oh-my-claude-sisyphus`), and a Claude Code plugin. It manages autonomous coding pipelines (autopilot, ralph), multi-worker team sessions via tmux, skill learning and injection, and provides extensive tooling including LSP integration, AST analysis, Python REPL, and git-based checkpoints. Architecturally, OMC is a modular monolith with clear separation: a CLI entry point (Commander-based), a hook registry with ~20+ lifecycle hooks (session-end, pre-tool-use, persistent-mode, etc.), a team subsystem with state machines and tmux-based worker management, a HUD rendering engine, and a standalone MCP server exposing 33+ tools. The system integrates with multiple AI providers (Claude, Codex, Gemini, Grok, Cursor) and git hosting providers (GitHub, GitLab, Azure DevOps, Bitbucket, Gitea). External dependencies include `@anthropic-ai/claude-agent-sdk`, `@modelcontextprotocol/sdk`, `better-sqlite3`, `zod`, and `commander`. Key data flows include: user input β CLI or Claude Code hook β hook dispatcher β feature modules (autopilot, team, skills) β tool execution β state persistence (filesystem, SQLite). The system maintains extensive state on disk (`.omc/state/`, session transcripts, team events, skill files) and uses file-based locking for concurrency safety. The JEV shadow judgment system provides non-blocking evaluation of heuristic decisions against an external API for quality monitoring. The codebase demonstrates strong engineering practices: extensive test coverage (~3000+ tests), modular architecture with clear interfaces, atomic file operations, and security-conscious utilities (SSRF guard, token redaction, path traversal prevention). It is actively maintained with frequent releases and responsive issue triage.
Ruflo is a large, modular AI agent orchestration platform that extends Claude Code with a plugin ecosystem, swarm coordination, memory systems, neural learning, security enforcement, and federation. The core is a CLI tool (`npx ruflo`) that provides 26+ commands and 215+ MCP tools for managing agents, memory, security, hooks, and deployments. It includes a console UI inside Claude Code, a Svelte-based chat UI (ruvocal), and a React-based goal UI. The architecture is ADR-driven with clean separation into domain, application, and infrastructure layers across 35+ plugins and 20+ core packages. The system implements advanced patterns: CQRS for memory and swarm, circuit breakers for resilience, capability envelopes for security, Ed25519 signing for release integrity, and a harness loop for automated champion selection. It supports multi-model routing, hybrid vector search (HNSW + SQLite), watermarking (WASM-based Gumbel scheme), and neural learning (PPO, DQN, Decision Transformer). Federation enables cross-installation agent collaboration with zero-trust, HMAC-signed envelopes, and PII-gated data flow. While the codebase is well-structured and actively maintained, it has some maintenance risks: duplicate helper scripts across multiple directories, hardcoded version numbers in smoke tests, experimental alpha plugins, and a large regression test suite that requires ongoing upkeep. The plugin ecosystem is extensive but some plugins may be aspirational rather than production-ready.
OpenRath is a framework for building LLM-powered AI agent applications with a torch-like API design. It centers on the Session abstraction (analogous to a Tensor) that flows through Sandboxes (execution environments), Memories (persistent state), and Tools (callable operations), orchestrated by Agents and Workflows. The framework provides session lineage tracking, sandboxed tool execution, memory persistence, and LLM provider abstraction, enabling the creation of multi-agent workflows with durable state and traceable provenance. The architecture is modular and extensible, featuring pluggable backends for sandboxes (local, OpenSandbox) and memory systems (local, OpenViking), a registry pattern for dynamic backend loading, and lazy evaluation of sessions for performance. Key innovations include write-ahead logging for crash-resistant session persistence, resource-keyed parallelism for safe concurrent tool execution, and unified garbage collection across session, sandbox, and memory domains. The framework is designed for production use, with extensive testing, clear documentation, and active maintenance.
DeepTutor is a full-stack AI-powered learning companion that provides chat, problem-solving, research, visualization, and structured learning capabilities. The system features a modular architecture with a Python/FastAPI backend, Next.js/React frontend, and a CLI interface. It implements an agent-based system where capabilities like deep solve, deep question, and deep research are implemented as loop capabilities with access to tools such as web search, code execution, and knowledge base retrieval. The platform includes a three-layer memory system for contextual awareness, supports multiple RAG backends (LlamaIndex, PageIndex, GraphRAG, LightRAG, IMA), and enables integration with external knowledge bases (Obsidian, MarginNote 4) and partner agents (Claude Code, Codex). The application employs a plugin architecture for extensibility, allowing new capabilities, tools, and subagents to be added via entry points. Configuration is managed through JSON/YAML files with runtime overrides, and services are decoupled via provider abstractions for LLMs, embeddings, search, and media generation. Real-time communication is facilitated by WebSockets and an internal event bus (StreamBus). The system supports multi-user deployments with role-based access control, isolated workspaces, and grant-based resource sharing for skills, knowledge bases, and partners. Deployment options include local installation, Docker containers, and CLI-only mode. The codebase demonstrates strong engineering practices with extensive unit testing, clear separation of concerns, and reusable patterns such as the capability protocol, tool composition, and atomic file operations. While the system's feature richness introduces complexity, its modular design allows for selective adoption of components.
This repository is a comprehensive educational resource focused on production-grade Generative AI systems, specifically designed for interview preparation and professional reference. Rather than being an executable application, it serves as a detailed knowledge base covering 16 critical areas of modern GenAI deployment, from foundational concepts like RAG systems and embedding technologies to advanced topics including multi-agent architectures, cost optimization, and system design. The content is organized into difficulty-levelled sections (Fresh, Intermediate, Advanced, Expert) with each topic featuring expected answers, common pitfalls (Red Flags), and real-world production statistics drawn from MENA enterprise deployments. What distinguishes this resource is its practical orientation - it emphasizes what actually works in production environments rather than theoretical concepts, with specific attention to trade-offs, failure modes, and optimization strategies observed in real implementations. While the repository contains no executable code and thus has no traditional software architecture or reusable components, it excels as structured educational content. The markdown-based organization allows for easy navigation, and the inclusion of SVG diagrams helps visualize complex architectures. The resource demonstrates particular strength in its coverage of production-specific concerns: evaluation methodologies, observability practices, security guardrails, and cost optimization techniques that are often overlooked in more theoretical treatments of GenAI topics.
ε°θ΄ (xiaobei) is an AI-powered multi-agent content marketing automation platform for small and medium businesses, built on top of the OpenClaw agent framework. It focuses on producing, publishing, and tracking content across major Chinese social media platforms (WeChat, Douyin, Xiaohongshu, Bilibili, etc.). The system uses a custom forked browser automation CLI (camoufox-cli) for platform interaction, a relay service for API signing, and SQLite databases for content and performance metrics. It includes multiple specialized agent crews (main, IT engineer, content producer, sales-cs) that collaborate through a modular skill system with thin wrapper scripts for agent invocation. The codebase demonstrates solid architecture with consistent patterns (DNA style analysis, workflow steps, patch application), but is heavily coupled to specific platform APIs and DOM structures, making it fragile to upstream changes. It is an actively developed, production-grade application aimed at automating social media operations.
OpenTeams is a multi-agent AI coding assistant platform that enables users to collaborate with multiple AI coding agents (Claude Code, Codex, Gemini CLI, etc.) in a single shared session. The system provides a local-first desktop application built with Tauri (Rust backend + React/TypeScript frontend), a standalone CLI tool (`openteams-cli`) written in TypeScript with Bun, and a web frontend. The backend is an Axum HTTP server in Rust with SQLite persistence managed via SQLx, supporting chat sessions, workflow orchestration, source control integration, approval systems, and a skill registry. The architecture is modular: a `crates` workspace separates database models, services, executors, deployment, and utilities, while the CLI uses Effect-TS for dependency injection and includes a TUI built with SolidJS. In production, the system orchestrates AI agents through the Agent Client Protocol (ACP) and provider-specific SDKs, managing execution processes, worktrees, and chat message queues. Key data flows include: user input β chat runner β agent execution (with tool calls, file edits, git operations) β event streaming back to frontend/CLI. The platform integrates with GitHub for PR review, issue tracking, and work item management, and supports custom MCP servers and skill plugins. External dependencies include SQLite, Tauri, Bun, various AI provider APIs (Anthropic, OpenAI, Gemini, Copilot), and GitHub CLI for certain operations.
The google/adk-go repository is a production-grade, open-source Go framework for building, orchestrating, and deploying multi-agent AI systems. It provides a comprehensive set of abstractions including agents, workflows, tool integrations, model backends (Gemini, OpenAI, Apigee), session management, artifact storage, and human-in-the-loop (HITL) support. The architecture centers around a workflow engine that supports graph-based execution with nodes, edges, retry, concurrency, and dynamic branching. It also implements agent-to-agent (A2A) and Model Context Protocol (MCP) for inter-agent communication. The framework is designed to be deployed via REST API, WebSocket, SSE, or as a console application, and includes launchers for Google Cloud Run and Vertex AI Agent Engine. While primarily an application framework, it also includes telemetry and analytics plugins for observability. The codebase is well-structured with extensive unit and integration tests, but some internal packages and experimental features (e.g., OpenAI model adapter) indicate ongoing development and potential for interface drift.
This repository implements a multi-agent AI system for generating comprehensive company research reports. The backend is a FastAPI application that orchestrates a LangGraph-based pipeline of specialized research agents: a grounding node that optionally crawls the company's website, parallel research nodes (company, industry, financial, news) that use Tavily search and LLM-generated queries to gather data, a curator that filters documents by relevance score, an enricher that fetches raw content, a briefing node that generates category-specific summaries using Gemini, and an editor that compiles the final report using GPT-4o. The frontend is a React application with TypeScript and Tailwind CSS that communicates via SSE streaming. The system is primarily a web application with a modular agent pipeline, not a data analysis project.
MultiGen is an AI agent platform designed for private, on-premise deployment. It features a planner-react architecture where a planner agent creates task plans and a react agent executes them using various tools (web browsing, shell commands, file operations, multimodal generation). The system maintains persistent session history via PostgreSQL, enabling conversation continuity and agent memory retention. Extensibility is supported through Model Context Protocol (MCP) and Agent-to-Agent (A2A) integrations. The backend is built with FastAPI using a layered architecture (Interface/Application/Domain/Infrastructure), the frontend with Next.js 14 App Router, and execution occurs in a sandboxed Docker container with Ubuntu, Chrome, and VNC. Real-time updates are delivered via Server-Sent Events (SSE) powered by Redis streams.
OASIS is a Python framework for simulating social media platforms (Twitter, Reddit) populated by LLM-powered agents. It is designed for computational social science research, specifically studying information spread, opinion dynamics, group polarization, and the impact of platform algorithms. The system is built on top of the CAMEL AI framework for LLM interaction. The architecture follows a modular, event-driven pattern. The core simulation loop is managed by OasisEnv, which coordinates agents (SocialAgent), a platform (Platform), an async message channel (Channel), and a social graph (AgentGraph). Actions are generated by agents (either manually or via LLM calls), passed through the channel, and executed by the platform against a SQLite database. A recommendation system (Recsys) provides personalized content feeds using various strategies. The codebase is a hybrid of a software library and a research project. While the core framework is well-structured with clean abstractions (e.g., Channel, AgentGraph, Recsys), it carries significant research code debt: tight coupling to the fast-moving camel-ai framework, global state in recommendation modules, skipped tests, hardcoded paths in experiments, and mixed sync/async patterns. The extensive YAML-driven experiment configurations and visualization scripts are tightly coupled to specific research questions (misinformation, polarization) and datasets. Maintenance is active in the context of ongoing research, but the codebase shows signs of technical drift (skipped tests, commented-out features, dependency on specific LLM API patterns). It is a powerful tool for researchers but would require significant hardening for production use.
AgentFlow is a framework for building and training agentic LLM systems using reinforcement learning, specifically Flow-GRPO (Flow-based Group Refined Policy Optimization). It provides a modular architecture with four core rolesβPlanner, Executor, Verifier, and Generatorβthat coordinate via evolving memory and tool calls over multiple turns. The system supports both inference (via a Solver) and distributed RL training (via a Trainer/Client/Server architecture built on verl, vLLM, and Ray). It includes a multi-provider LLM engine abstraction (OpenAI, Anthropic, Gemini, DeepSeek, etc.), a tool system with metadata-driven discovery, and instrumentation for tracing and reward collection. The repository is heavily research-oriented, with benchmark datasets (AIME24, GPQA, GAIA, HotPotQA, etc.) and training pipelines for optimizing agent planners. While primarily an application framework, it has strong data/research components, making it a hybrid project with a focus on reproducible agentic reasoning experiments.
Solace Agent Mesh is a production-grade, full-stack platform for building, deploying, and managing event-driven multi-agent AI systems. It comprises a Python backend (FastAPI, Google ADK, SAC), a React/TypeScript web UI (chat, projects, workflows, prompts, scheduled tasks), a Click-based CLI, and a Docusaurus documentation site. The system uses a microservices-like architecture with separate gateway (HTTP/SSE), platform service (model configuration, storage), and agent components, all communicating via the Solace event broker and the A2A (Agent-to-Agent) protocol. Key entry points include the CLI (`solace-agent-mesh`), the web UI (`/chat`, `/projects`, `/workflows`), and the REST API (`/api/v1`). Data flows involve user requests via SSE, agent orchestration with workflow DAGs, tool execution (built-in, MCP, OpenAPI), artifact storage (S3/GCS/Azure/filesystem), and session persistence in SQL databases (SQLite/PostgreSQL/MySQL). External dependencies include a Solace broker, LLM endpoints (LiteLLM), optional databases, and cloud storage services. The repository is marked as deprecated (Development Status 7 β Inactive), with the new version being a desktop application.
This repository implements a Chinese-localized version of the Stanford Generative Agents simulation. It creates a virtual town with 25 AI agents powered by LLMs (primarily via Ollama) that autonomously organize events, form relationships, and exhibit human-like behaviors. The system consists of a simulation engine that runs agent interactions over time, a compression tool to generate replay data, and a Flask-based web frontend for visualizing the simulation. Agents maintain memory, generate plans, reflect on experiences, and interact through LLM-driven conversations and actions. The application supports local deployment with configurable LLMs and includes features like breakpoint resumption and conversation logging.
KEA Research is a full-stack multi-provider AI research chat application. The backend is a FastAPI service backed by SQLite/SQLAlchemy, exposing chat, admin, health, user, and TTS endpoints. The frontend is an Astro + TypeScript single-page app that stores chat history and pipeline data in IndexedDB, with a Bootstrap-based UI. Deployment is containerized via Docker Compose with nginx handling TLS and SSE-friendly proxying. The core product logic is a four-step research pipeline: initial multi-provider responses, Mixture-of-Agents refinement, peer evaluation, and a final KEA synthesis with facts, consensus, flagged claims, rankings, and confidence scores. A web admin panel manages providers, models, provider sets, users, and app settings, while the frontend supports research layers, bookmarks, notes, drafts, export, attachment ingestion, and text-to-speech via a Kokoro ONNX model.