Castor: A Self-Hosted AI Agent for Business Workflows
Castor is an open-source, self-hosted AI agent designed for business operations like customer service, internal automation, and knowledge retrieval. It supports various OpenAI-compatible LLMs, including local models, and emphasizes data privacy by keeping data on the user's infrastructure.
HarnessClaw Engine: A Go-based LLM Programming Assistant with WebSocket and Tool Calling
HarnessClaw Engine is an open-source LLM programming assistant engine built with Go. It supports WebSocket for multi-turn dialogues, tool calling, permission control, and skill extension, and recently released version v0.0.18.
Aguara: Local-First Security Scanner for AI Agents and Software Supply Chains
Aguara is an open-source, local-first security scanner designed to detect vulnerabilities in AI agents and software supply chains. It identifies risks such as prompt injection, tool poisoning, unsafe GitHub Actions, secret exfiltration, and compromised packages across various ecosystems without relying on SaaS or LLM calls.
PentesterFlow/agent: AI-Powered Offensive Security CLI Tool
PentesterFlow/agent is an open-source, agentic AI command-line interface (CLI) designed for offensive security tasks. It assists penetration testers and bug hunters with recon, enumeration, validation, evidence collection, and reporting, while maintaining human oversight.
Wide-Moat's Open-Source MCP Server for LLM-Powered Computing
Wide-Moat has released an open-source MCP server designed to provide Large Language Models (LLMs) with their own managed computing environments. This self-hosted solution offers Docker workspaces with integrated browser, terminal, and code execution capabilities, enabling LLMs to perform complex tasks autonomously.
Antfly: A Distributed Search Engine for Multimodal AI Data
Antfly is an open-source, distributed search engine built in Zig, designed for multimodal AI data. It integrates full-text search, vector similarity, and graph traversal, with built-in RAG agents and support for various AI models and deployment environments.
Ataraxy-Labs Introduces Sem: Semantic Version Control for Coding Agents
Ataraxy-Labs has released `sem`, a semantic version control tool built on Git that provides entity-level diffs, blame, and impact analysis. Designed for coding agents, it supports 26 programming languages via tree-sitter and integrates directly with Git workflows.
StatsPAI: Agent-Native Python Platform for Causal Inference and Econometrics
StatsPAI is an open-source Python platform designed for causal inference and applied econometrics, featuring a unified API, extensive method coverage, structured result objects, and machine-readable schemas. It is built with AI agents in mind, providing discovery metadata for its functions and validation statuses for certified numerical evidence.
HotPlex: A Unified Access Layer for AI Coding Agents
HotPlex is an open-source Go gateway designed to provide a unified WebSocket interface for various AI coding agents. It supports multiple platforms like Web, Slack, and Feishu, and includes features such as deterministic sessions, AI-native cron scheduling, and an embedded web chat with an admin UI.
Open-Source-Legal's 'cite' Project: Version Control for Knowledge and AI Agents
Open-Source-Legal has developed 'cite', a Python-based GitHub project designed to create a "ground truth layer" and version control system for knowledge, facilitating collaboration between humans and AI agents. It aims to build an open citation graph for document repositories, allowing agents to traverse relationships between documents and propose new annotations.
Thunderbolt: An Open-Source, Cross-Platform AI Client for On-Premise Deployment
Thunderbolt is an open-source, cross-platform AI client developed by Thunderbird, designed for on-premise deployment. It emphasizes user control over models and data, aiming to eliminate vendor lock-in. The project is currently in active development, targeting enterprise customers.
Strands Agents Tools: A Python Library for AI Agent Capabilities
Strands Agents Tools is a Python library that provides a collection of tools designed to enhance the capabilities of AI agents. It offers functionalities for file operations, shell integration, memory management, web interactions, API requests, Python code execution, mathematical operations, and integrations with AWS, image/video/audio processing, and more.
AgentCore: A Minimal Go Library for Building AI Agent Applications
AgentCore is a Go library designed for building AI agent applications, emphasizing a minimal and composable core with extensibility. It supports single and multi-agent architectures, offering features like tool call gating, sub-agent invocation, context management with auto-summarization, and a unified event stream for lifecycle events.
Awesome AI Agents 2026: A Curated List of AI Agent Resources
The `awesome-ai-agents-2026` GitHub repository provides a curated list of AI agent frameworks, tools, platforms, and resources. It aims to be a definitive guide for the year 2026, which the repository's description labels as the year AI agents went mainstream.
DocETL: An Agentic LLM-Powered System for Data Processing and ETL
DocETL is a GitHub project that provides an agentic LLM-powered system for data processing and Extract, Transform, Load (ETL) operations. It is designed to handle unstructured data analysis and semantic data processing.
Anthropic Announces Claude Fable 5
Anthropic announced Claude Fable 5 alongside Claude Mythos 5 with expanded capabilities for coding, knowledge-intensive tasks, and practical developer workflows. The announcement is also reflected in the OpenRouter model listing, reinforcing that the model is now visible across broader distribution surfaces. This launch indicates a continuation of rapid model iteration focused on production-grade AI application support.
OpenAI Enhances GPT-Rosalind with Advanced Life Sciences Capabilities
OpenAI has introduced new capabilities for GPT-Rosalind, a model designed to support life sciences research. These enhancements include improved biological reasoning, medicinal chemistry expertise, genomics analysis, and experimental workflow functionalities.
Cohere Introduces North Mini Code: A New Model Tailored for Developers
Cohere has announced North Mini Code, its first model specifically designed for developers. This new model aims to provide enhanced capabilities for coding tasks and development workflows.
PAR3D: A Unified 3D-MLLM for Part-Aware Scene Understanding
Researchers have introduced PAR3D, a unified 3D Multimodal Large Language Model (3D-MLLM) framework designed to enhance 3D scene understanding by focusing on fine-grained part structures in addition to objects. This approach aims to improve embodied interaction with 3D environments.
Her: A Detective for Claude Code Sessions
Her is a new tool designed to act as a detective for Claude code sessions. It aims to help developers understand and debug their interactions with Claude, making the process more transparent and manageable.
Causally Evaluating the Learnability of Formal Language Tasks
Researchers propose a new methodology for evaluating the learnability of tasks in language models, moving beyond standard correlational analysis. By using formal languages derived from probabilistic finite automata, they introduce the 'binning semiring' to causally control data frequency and measure learnability. This approach aims to address the inherent flaws in correlational evaluations, which can lead to incorrect conclusions.
Activation-Based Active Learning for In-Context Learning: Challenges and Insights
A new research paper investigates the use of transformer model activations for selecting in-context examples in large language models. The study, which includes a comprehensive analysis using Llama-3.2-3B and Qwen2.5-3B, found that MLP outputs based on massive activations or statistical moments do not correlate with example quality or task performance, suggesting that this approach is not effective for in-context learning.
GPT-2 Based Text Generation Model by rupeshpanda on Hugging Face
A new text generation model, `rupeshpanda/gita-text-generation-gpt2`, has been identified on Hugging Face. Built on the GPT-2 architecture and utilizing the Transformers library, this model is designed for text generation tasks.
NexusBench-trajectories Dataset Released by AgentSuite
AgentSuite has made the NexusBench-trajectories dataset publicly available on Hugging Face. This dataset contains per-model agent trajectory data for the NexusBench benchmark, offering detailed insights into agent behavior across various tasks.