Open Source Security Projects
Discover 166 open source Security repositories from GitHub, each with an analysis of what it does, key features, use cases and alternatives. Security projects here are most often combined with Self Hosted, Python and AI Agents. Last updated October 4, 2026.
166 repositories · updated October 4, 2026
agent-governance-toolkit: Govern Autonomous AI Agent Actions
Microsoft’s Agent Governance Toolkit adds policy checks, identity, audit logging, sandboxing, and reliability controls around autonomous agents. It suits teams integrating agent frameworks that need application-level enforcement and evidence of tool-call decisions.
borgbackupserver: Manage Borg Backups Across Endpoints
Borg Backup Server is a self-hosted web app for scheduling and managing Borg backups across multiple machines. It suits people who want a central interface for backup jobs, restores, and repository operations without opening inbound connections to their endpoints.
toolsdk-mcp-registry: Discover and Run MCP Servers
ToolSDK MCP Registry catalogs MCP servers and provides a self-hosted gateway for searching and executing them. It is aimed at teams building AI applications that need centralized tool discovery, protocol bridging, and controlled execution.
agent-sandbox: Run Coding Agents in a Locked-Down Environment
Agent Sandbox runs AI coding agents in Docker containers with restricted workspace access and default-deny network controls. It suits developers who want to limit an agent’s access to local files, network services, and credentials while retaining CLI or IDE workflows.
anythingmcp: Turn APIs and Databases into MCP Tools
AnythingMCP is a self-hosted gateway that turns REST, SOAP, GraphQL, OData, and SQL systems into tools for MCP-compatible AI clients. It combines connector generation with access controls and auditing for teams that want AI access to existing systems.
gh-aw-firewall: Restrict Network Access for Agentic Workflows
A Docker-based firewall for running agentic workflow commands with outbound traffic limited to approved domains. It also isolates LLM credentials in a sidecar, making it useful for teams that need tighter network and secret controls around automated agents.
awesome-ai-agent-attacks: Track Documented AI Agent Security Incidents
A sourced timeline of real AI agent security incidents, breaches, and vulnerabilities from 2024 to 2026. Useful for security researchers, incident responders, and teams assessing risks in agentic systems.
ADR: Secure and Monitor Enterprise AI Agents
ADR is an enterprise security toolkit for discovering AI tools, collecting agent activity, benchmarking defenses, and detecting risky behavior. It is aimed at security teams evaluating or monitoring AI agents across employee endpoints and customer-facing systems.
start-technologies: Run a Self-Hosted Server and Home Network Stack
Start Technologies is a monorepo for StartOS and related tools for self-hosting, service packaging, registries, tunneling, and router management. It suits people who want to run services on hardware they control and developers building for the Start9 ecosystem.
OpenSandbox: Run AI Agents in Isolated Sandboxes
OpenSandbox provides isolated environments where AI applications can run code, commands, and browser or desktop tasks. It supports local Docker use and Kubernetes deployment through a shared sandbox API.
mercury-agent: Run a Permission-Aware AI Agent Across Channels
Mercury is a self-hosted AI agent for CLI, Telegram, and web use, with permission prompts, persistent memory, token budgets, and scheduled tasks. It suits people who want an always-on assistant they can configure and operate locally.
CubeSandbox: Run Secure, Fast Sandboxes for AI Agents
CubeSandbox is a self-hostable sandbox service for running untrusted AI-agent workloads in hardware-isolated microVMs. It combines E2B SDK compatibility with snapshots, network controls, and single-node or multi-node deployment.