Search

Pisces is an AI assisted, self-healing security patcher, A CLI tool for Python codebases. It supports HITL. It doesn't just find vulnerabilities — it fixes them, validates the fixed patch against your test suite, and rolls back automatically if anything breaks. https://pypi.org/project/pisces-cli/

A highly resilient, autonomous LLM agent built for Google Apps Script. Features granular RBAC, hierarchical method blocking, secure network sandboxing, and intelligent middleware (auto-pagination & latency resolution) to execute complex Google API workflows safely via MCP schema discovery.

openinterpreter is an open-source AI developer tool hosted on GitHub. A lightweight coding agent for open models like Deepseek, Kimi, and Qwen OpenTools lists it for builders who need practical AI infrastructure, agent, security, caching, or model-workflow components they can inspect and test before adoption.

Multi-Agent AI Code Auditor is a Python-based code review system that uses AST parsing, Bandit, Radon, and LLM-based agents to analyze code for security, quality, and complexity. The findings are summarized into a final report, saved as JSON, and benchmarked across multiple Python programs.

finsightAI is a modular, multi-agent financial assistant powered by Google's Agent Development Kit (ADK) and Gemini LLMs. It delivers tailored financial insights using real-time user context sourced via a secure FI MCP (Financial Information Multi-Context Platform) server.

AgentStack is a production-grade multi-agent framework built on Mastra, delivering 50+ enterprise tools, 25+ specialized agents, and A2A/MCP orchestration for scalable AI systems. Focuses on financial intelligence, RAG pipelines, observability, and secure governance. ACP Openclaw, Gemini CLI, Opencode

The Fraud Investigator Assistant shows how AWS Bedrock, MCP servers, and Strands expert agents can transform fraud investigations. By combining private MCP servers for secure data with open-source MCPs for public intelligence, it speeds up case analysis and ensures compliance, reducing investigation timelines significantly.

Stateful Telegram AI Assistant powered by Gemini 2.5 Flash & LangChain. Securely manages Google Calendar (CRUD) via custom tools. Features SQL-backed conversational memory for dynamic chats, zero-downtime error handling, strict access control, and cloud-ready architecture.

MCP server that teaches AI assistants Magento 2 coding standards — validate code, look up correct patterns, check security, and apply theme-specific rules (Hyva, Luma, Breeze, Porto). Works with Claude, Cursor, Gemini CLI, VS Code Copilot, and any MCP-compatible client.

Full-stack air quality analytics platform built with FastAPI, React, and MySQL. Aggregates multi-source PM2.5/PM10 data, performs multi-city comparison and time-series forecasting (SARIMAX), and integrates an LLM-based planning agent with tiered access, secure APIs, and PDF reporting.

Seven agentic AI projects — LangChain SQL agent, LangGraph routing, CrewAI & AG2 multi-agent apps, the full BeeAI progression, and Model Context Protocol client/server + security stack — each with verified tests (73 total), incl. real end-to-end MCP protocol integration tests.

Minimal, secure Google Workspace MCP server — 43 tools across 10 services (Gmail, Calendar, Drive, Docs, Sheets, Slides, Forms, Tasks, Chat, Meet). Only official Google libraries, zero third-party wrappers, fully auditable. Works with Claude Code, Kiro, Cursor, Windsurf, Codex.

As autonomous AI agents are increasingly deployed across enterprise workflows, their tool-calling capabilities (e.g., executing shell commands, sending emails, or modifying databases) expose critical security vulnerabilities. Prompt injections or reasoning loops can lead to catastrophic unauthorized actions—such as dropping database tables or leaki

Agent orchestration platform & operating-system-workspace. It powers Agents, mini apps, documents, CRM, CMS, chat, workflows, tools, and secure browser access across desktop, mobile, and web using skills and MCP. Built on Aurora Cloud, it comes bundled with production ready agents, agentic runtime, MCP + MORE.

Salesforce AI Assistant — An intelligent, LLM-powered assistant that enables natural language interaction with Salesforce metadata and data. Built with LangGraph for agent workflows, Weaviate for semantic search, MongoDB for short-term memory, and high-quality embedding models. Supports SOQL generation, metadata exploration, and secure multi-org

Framework for building and managing multi-agent systems with Model Context Protocol (MCP) and LangGraph support. Features a modern React UI, JWT-secured agent communication, and dynamic LLM provider integration (OpenAI, Azure, Google). Easily create, configure, and monitor agents that connect to external tools—all from a single interface.

NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest compute efficiency and accuracy for developers to build specialized agentic AI systems. The model is fully open with open-weights, datasets and recipes so developers can easily customize, optimize, and deploy the model on their infrastructure for maximum privacy and security.

Minimal, reproducible demonstration for the work "Agentic FIWARE: MCP and Guardrail Generic Enablers for Secure and Auditable Smart City Orchestration". It wires together an MCP server (FastAPI), FIWARE Orion (NGSI-v2), a simple policy engine that simulates OPA, and a host process that stands in for an LLM executing plans with audit-friendly trace

NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest compute efficiency and accuracy for developers to build specialized agentic AI systems. The model is fully open with open-weights, datasets and recipes so developers can easily customize, optimize, and deploy the model on their infrastructure for maximum privacy and security.

Aegis Agent Framework is an engineering-grade AI agent runtime built on Python 3.11+ and pydantic v2. Designed with an "offline-first" and secure-isolation philosophy, it features rigorous architectural boundaries and a 5-slot context engineering model. With out-of-the-box support for 11 model adapters and native integration with MCP and A2A protoc