Frameworks, tools, and infrastructure for building, deploying, and orchestrating AI agents — multi-agent systems, LLM coding assistants, MCP servers, memory layers, and orchestration platforms.
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web, then synthesizes a grounded summary. Watchlist mode for recurring research. #1 GitHub trending at 10,211 stars (+2,685/day). First breakout 'single-purpose agent skill' — not a framework.
Meta's self-referential self-improving agents — a meta agent modifies both the task agent AND its own improvement procedure. Cross-domain transfer (robotics strategies improving math grading). 121 HN pts. CC non-commercial license.
Autonomous CLI agent for deep financial research — task planning, self-reflection, real-time market data. TypeScript/Ink/LangChain. Self-validating with LLM-as-judge evaluation suite. 18,972 stars. Second vertical agent breakout after pentagi (security).
AI memory engine — persistent memory and knowledge management for agents. Accelerated from +344/day to +810/day (18,863 stars). Memory/synthesis layers becoming first-class demand centers.
Automated content creation and monetization pipeline — AI-generated short videos with multi-platform publishing. 25,139 stars (+3,006/day). Content automation as a distinct agent category.
ByteDance's SuperAgent harness — researches, codes, and creates using sandboxes, memories, tools, skills, and subagents. Accelerated to 48,267 stars (+2,394/day) by Mar 27, up from 36.6K debut on Mar 23. Fastest-growing multi-agent framework this week.
Fully autonomous AI pentesting system in Go — performs complex penetration testing tasks without human oversight. 12,400 stars (+1,069/day). Go implementation signals production readiness. Fastest-growing security-focused agent.
Claude Code skills that turn codebases into interactive knowledge graphs — shift from 'AI writes code' to 'AI understands code.' Multi-platform (works with Codex). 4,700 stars (+489/day).
48 AI agents mirroring real game studio hierarchy — 36 workflow skills with full coordination system. Most complex multi-agent coordination system documented on GitHub. 2,500 stars (+328/day).
Autonomous development loop for Claude Code — self-referential execution with exit detection. 8,066 stars. Part of the Claude Code ecosystem expansion beyond plugins into workflow automation.
Agent memory with SOTA local GPU retrieval — GPU-accelerated vector store for agent context. Part of the Mar 22 agent infrastructure gold rush (4 independent memory/context tools in one day).
Compress codebase context for LLMs by 78% using Rust — reduces context window consumption without lossy compression. Complementary to memory tools: compress first, then store/retrieve.
Agent-vs-agent security testing — 3 Claude agents (Red vs Blue) hack and patch code autonomously. Novel paradigm: using AI agents to red-team code produced by other AI agents before deployment.
Agentic Linux kernel code review system — AI-powered review of kernel patches. First domain-specific agent for OS kernel development.
Pure C/Metal inference engine running Qwen3.5-397B (397 billion parameters) on a MacBook Pro with 48GB RAM at 4.4+ tokens/sec. No Python, no frameworks — just C and hand-tuned Metal shaders. The entire 209GB model streams from SSD.
Multi-agent LLM financial trading framework — agents analyze market data, execute trades, manage risk. 41,343 stars (+1,760/day), with Chinese adaptation TradingAgents-CN at 21K. Domain-specific multi-agent validated.
Meta-prompting, context engineering, and spec-driven development system for Claude Code — competing with superpowers for the 'canonical agentic methodology' position. 39,200 stars, sustained at +301/day.
Agent harness performance optimization system — 13 expert agents, 40+ skills, 32 commands. By Anthropic hackathon winner. Crossed 100K milestone (~108,000 stars). Second Claude Code ecosystem project (after superpowers) to hit 100K.
MoonshotAI/Kimi's drop-in replacement for standard residual connections in Transformers — depth-wise attention over previous layers. Achieves same performance with 1.25x less compute. 236 HN pts. Published at ICLR 2026.
LangChain's open-source framework for building internal coding agents — triggers from Slack/Linear/GitHub, runs in isolated cloud sandboxes, auto-creates PRs. Sustained growth past initial spike.
Three open-source TTS models — smallest under 25MB, CPU-only, ONNX-based. Sustained to 554 HN pts (Day 3+). Sub-25MB footprint enables bundling with CLI agents for offline voice. Viable piper-tts alternative.
Autoresearch for SAT solvers — AI-driven research agent that autonomously improves SAT solver performance. Karpathy's overnight-experiment paradigm applied beyond ML.
Andrew Ng's curated, versioned API documentation for coding agents — agents search, fetch, and annotate docs with a self-improvement feedback loop. 11,465 stars (+5,224/week), sustained growth.
Nvidia's orchestration framework for secure sandboxed agent deployment — policy-driven containment with hot-reloadable security policies and Nemotron inference integration. 15,400 stars (+140/day), sustained.
Autonomous wildfire tracking from satellite and weather data — AI agents applied to environmental monitoring. First non-software agent application in tracking.
Open-source agent for formal proof engineering using Lean 4 — highest-engagement formal methods story ever on HN. AI-native theorem proving enters the mainstream.
Official LangChain agent harness with planning, web browsing, and coding built on LangGraph — 16,621 stars (+5,487/week), accelerating. In direct competition with superpowers (shell-native) and Claude Code Skills (IDE-native).
Reusable middleware for production AI agent deployment — addresses the gap between agent prototyping and production-grade lifecycle management. ArXiv paper.
Distributed multimodal search, memory, and graph database in Go — agent-native data infrastructure combining vector search with knowledge graphs.
YC W26 — maps and geospatial API designed specifically for AI agent consumption. First agent-native geospatial infrastructure.
Agent memory engine — build and manage knowledge graphs for AI agents. 14,047 stars with sustained +270/day growth. Fills the persistent memory layer in the agent stack.
Agentic OS for physical space — 'vibecode humanoids.' First robotics/physical-space agent framework on GitHub trending. 966 stars.
Vault for AI agents in Rust — secure credential management for agent workflows. Addresses the #1 agent security gap with compiled-language security.
Claude Code session analytics — first dedicated observability tool for Claude Code sessions. Track costs, usage patterns, and session performance.
ByteDance's open-source context database for AI agents — replaces fragmented vector stores with a file-system paradigm unifying memory, resources, and skills under a viking:// protocol. Self-evolving memory with built-in iteration loops. 18,100 stars (+151/day), sustained.
HKU's 'Making ALL Software Agent-Native' — auto-generates CLI wrappers for any software (Blender, LibreOffice, FFmpeg) via a 7-phase pipeline, with built-in --json for machine consumption. 12,800 stars.
Karpathy's autonomous ML research framework — 630-line script that runs experiments overnight on a single GPU. 333 experiments unsupervised in one night. Shopify CEO reported 19% validation improvement. 29,100 stars in one week.
Agentic proxy that compresses agent context before it hits the LLM — sits between your AI agent and the API, compacting history in the background so agents never wait. Addresses the core context engineering bottleneck.
Open standard for git-native AI agents — your repo IS the agent. Framework-agnostic, works with Claude/OpenAI/CrewAI. Version agents with git branches, test behavior in CI, share via public registry.
Minimal 12MB single-binary AI agent framework — the anti-complexity counter-movement to agent framework bloat. No dependencies, no framework lock-in.
Comprehensive RAG platform combining Langflow, Docling, and Opensearch. 3,317 stars (+2,537/week), sustained growth.
Backend built for agentic development — give agents everything they need to ship fullstack apps. Agent-native infrastructure. 3,951 stars, +477/day Day 2 acceleration.
Agent memory that learns — adaptive memory improving over time rather than static storage. From vectorize-io. 5,009 stars (+2,183/week), sustained growth.
Official Anthropic-managed directory of high-quality Claude Code plugins. Curated ecosystem vs open marketplace. 11,069 stars, steady +411/day.
Agentic skills framework and software development methodology — sustained growth to ~115,000 stars. One of fewer than 70 GitHub repos to ever hit 100K. Official Anthropic marketplace acceptance.
IM chatbot infrastructure platform — multi-platform bot framework for messaging services. 22,700 stars with sustained +342/day growth.
Formal behavioral contracts for AI agents — Design by Contract pattern applied to agent systems. Exited trending on Day 4 as predicted. ~1,250 stars.
Pre-built agent marketplace — shell-based agent configurations for instant deployment. Surged to 54,556 stars (+25,297/week), the single highest weekly star gain in 2026 tracking. Still trending on Mar 19.
Chinese-origin multi-agent sentiment analysis framework — 37,654 stars. Part of the 666ghj swarm intelligence ecosystem.
NousResearch's adaptive learning agent — 'the agent that grows with you.' 12,835 stars (+1,278/day), sustained adoption beyond initial hype.
Lightweight authorship assertion protocol — a human.json file (like robots.txt) for declaring content authorship with a vouch mechanism.
Chinese-origin swarm intelligence engine — 'predict anything' using particle swarm and ant colony algorithms. Sustained velocity pushing to 39,270 stars (+15,056/week). No deceleration after two weeks.
Alibaba's JavaScript in-page GUI agent — control web interfaces with natural language. 11,532 stars (+6,794/week), sustained growth.
YC W26 — 'Vercel for filesystem-based agents.' Deploy agent workflows by pushing to Git with built-in observability.
Community response to Moltbook crisis — alternative collaborative coding notebook.
Microsoft's high-velocity agent framework — 12,951 GitHub stars.
Persistent memory layer for Claude — automatic session capture. 37,126 stars with sustained +1,045/day growth, one of the most-starred Claude Code plugins.
Benchmark for LLM long-context retrieval — find a fact in a large document.
AI orchestration framework for complex multi-step agent tasks.
MCP server providing up-to-date library documentation for AI coding agents.
Command-line interface for Moonshot AI's Kimi LLM models.
Unified multi-tier memory architecture for persistent AI agent context.
Collection of leaked system prompts from major AI assistants.
Email API for AI agents — YC S25 launch with 106 HN points.
Cloudflare self-hosted AI agent running on Cloudflare Workers.
Robust background job processing with queue management, now for Python.
Intercept LLM API traffic and visualize token usage in a real-time terminal dashboard. Track costs and debug prompts.
Persistent subconscious memory layer for Claude agents using Letta.
Curated prompt library for AI-assisted code review workflows.
SpecFlow integration for Claude Code — BDD test automation with AI.
Rapid application development framework with AI-assisted code generation.
Community-driven repository of verified, tested AI prompts.
Tmux + credit rotation wrapper for running multiple Claude Code sessions.
External orchestration layer for Claude Code — model-agnostic multi-agent coordination. Now rebranded as ruflo, surged to 25,424 stars (+1,397/day).
Inspect Claude's internal reasoning and tool calls before they execute.
MCP server for email — let AI agents send and read emails programmatically.
OpenAI's experimental lightweight multi-agent orchestration framework.
AI-assisted development environment with intelligent code analysis.
MCP URL toolkit — fetch and parse web content for AI agent context.
GPU-accelerated charting library for real-time high-performance data visualization.
TypeScript AI agent framework — workflows, memory, tool integrations, and evals.
AI-powered code generation tool by 21st.dev.
High-performance rendering library generated with AI assistance (1M+ lines of code).
Run large language models locally with a simple CLI — Llama, Mistral, and more.
Open-source AI coding agent built for the terminal — provider-agnostic alternative to Claude Code. Surged to 1,228 HN points (605 comments) on Mar 20. 126,000+ GitHub stars, 5M monthly active users. The open-source coding agent category has a clear leader.
Anthropic's open-source sandbox runtime for safe agent code execution.
Train LLMs with historical constraints to study knowledge cutoff effects.
Shared memory coordination for parallel AI coding agents with evidence-based commits.
Microsoft's multi-agent conversation framework for complex AI workflows.
Role-based multi-agent framework where AI agents collaborate like a specialized crew.
Popular framework for building LLM-powered applications and agent chains.
Deploy AI agents as REST APIs with auto-scaling infrastructure.
Standard protocol for agent-to-agent communication and interoperability.
Computer-use agent framework — let AI agents control desktop GUIs.
Human-in-the-loop controls for AI agents — require human approval before sensitive actions.
Real-time AI agent framework — build voice/video agents with LLMs.
Open standard connecting AI models to tools and data sources — protocol spec update.
AI infrastructure platform for running agent workloads at scale.
Agentic infrastructure platform — connect AI agents to 100+ APIs without integration code.
AI coding agent for large codebases — long-running multi-file tasks in the terminal.
Persistent memory store for Node.js AI agents.
Multi-agent builder with OpenAI tools — visual workflow editor for agent pipelines.
Framework for deploying AI agents as production services.