Agentos
Build and govern policy-compliant AI agents with safety enforcement and compliance checking via AgentOS over MCP.
Overview
io.github.imran-siddique/agentos is a MCP server for the Build phase that helps you build and manage policy-compliant AI agents with safety enforcement and compliance checking.
What is this MCP server?
- agentos-mcp-server npm stdio package v1.0.1 from agent-os repo
- Policy enforcement with AGENTOS_POLICY_MODE strict vs permissive
- Safety enforcement and compliance checking for agent behavior
- Configurable AGENTOS_LOG_LEVEL for debug through error
- npm package agentos-mcp-server version 1.0.1
- Policy modes: strict (block violations) or permissive (warn only) via AGENTOS_POLICY_MODE
Community signal: 70 GitHub stars.
What problem does it solve?
Custom agents can take destructive or non-compliant actions unless policy checks are wired outside the prompt.
Who is it for?
Solo builders productizing agents who need explicit policy modes and audit-friendly logging before customer-facing automation.
Skip if: Simple one-shot codegen with no autonomous tools, or teams unwilling to configure policy semantics and ops logging.
What do I get? / Deliverables
You get an MCP-managed layer that enforces AgentOS policies and surfaces compliance results while agents run.
- Policy-checked agent configurations managed through MCP
- Violation blocking or warnings per AGENTOS_POLICY_MODE
- Operational logs at the chosen AGENTOS_LOG_LEVEL
Recommended MCP Servers
Journey fit
Agent frameworks and guardrails are assembled during Build when you shape how autonomous tools behave in your product. Agent-tooling is the shelf for MCP servers that create, configure, or supervise agents—not generic CRUD APIs.
How it compares
Agent governance MCP from the AgentOS stack, not a lightweight single-integration utility server.
Common Questions / FAQ
Who is io.github.imran-siddique/agentos for?
Builders shipping agentic features who want AgentOS policy enforcement and compliance checks reachable from Claude Code, Cursor, or other MCP clients.
When should I use io.github.imran-siddique/agentos?
Use it while designing agent-tooling in Build, and again in Ship security when you need strict policy blocking before launch.
How do I add io.github.imran-siddique/agentos to my agent?
Install agentos-mcp-server via npm with stdio transport, set AGENTOS_POLICY_MODE and optional AGENTOS_LOG_LEVEL, and register the server in your MCP config.