MCP Server v1.0.1

AgentOS MCP Server v1.0.1

Model Context Protocol server for Claude Desktop integration.

Features

• 8 tools for agent lifecycle management, policy enforcement, and runtime governance
• 3 resources for policy templates, agent status, and governance dashboard
• 3 prompts for agent creation, policy review, and compliance audit

Install

npx agentos-mcp-server --stdio

Links

• Glama Listing
• Documentation

v2.0.1

What's Changed

Bug Fixes

• Fix broken \chat-agent\ example (imported non-existent \�gent_os.emk.EpisodicMemory, now uses local \memory.py)
• Fix broken \ ool-using-agent\ example (\ToolRegistry\ → \get_registry\ import)
• Suppress \google.generativeai\ FutureWarning that appeared on every import

Documentation

• Complete CONTRIBUTING.md with PR review process, release cadence, and code of conduct references
• Add requirements note for examples that need external dependencies

Full Changelog

v1.1.0...v2.0.1

v1.1.0 — Critical Governance Fixes

What's New

Critical Audit Fixes

• Drift detection implemented — post_execute() now checks drift_score against drift_threshold and emits POLICY_VIOLATION events on breach
• CrewAI tool interception fixed — pre_execute() result now enforced; raises PolicyViolationError when governance blocks a task
• Human approval enforcement — require_human_approval now checked in Anthropic and CrewAI adapters (was OpenAI-only)
• Type-safe comparisons — confidence and drift_score checks guard against non-numeric types

Features

• Tool execution via registry in OpenAI adapter
• Confidence threshold enforcement
• Awesome-list badges (AI-Agents, agent-skills, devops-mcp)

Tests

997 tests passing

Full Changelog: v1.0.0...v1.1.0

v1.0.0 - Agent OS Kernel

Agent OS v1.0.0 - Production-Ready AI Agent Governance Kernel

The foundational runtime for AI agent governance. This release includes:

Core Modules

• Ambient Authority (AMB) - Capability-based access control for AI agents
• Identity & Trust Protocol (IATP) - Zero-trust agent authentication
• Nexus - Multi-agent coordination and service mesh
• Policy Engine - Declarative governance policies with OPA integration
• Audit System - Immutable audit trails for compliance

Highlights

• 1,500+ tests with comprehensive coverage
• Python 3.9+ support
• Full async/await architecture
• Extensible plugin system
• Framework integrations (LangChain, CrewAI, AutoGen, Semantic Kernel)

Links

• Documentation
• PyPI
• Website

v1.3.2 — Workshop Ready + Framework Integrations

What's New in v1.3.2

Framework Integrations

• CrewAI adapter: Fixed module-level logger bug that caused NameError during agent execution
• 6 framework PRs actively maintained: LlamaIndex, Lightning AI, A2A, CrewAI, AutoGen, Semantic Kernel

Community & Workshops

• Created 82 workshop-ready issues for hands-on contributor workshops
• Added tech debt markers throughout codebase for Octane scanner demos
• 3 new labels: workshop, code-quality, beginner-friendly

Distribution

• Now showing PyPI badges with download counts on README
• pip install agent-os-kernel — version 1.3.1 on PyPI

Awesome List Presence

• Featured in awesome-llm-apps (93k★)
• 9 additional awesome list submissions pending review

v0.5.0

KernelSpace implementation with real PolicyEngine enforcement