Durable sessions
Append-only event log per session. Resume from anywhere; survive restarts and crashes.
OpenMA (Open Managed Agents) is the infrastructure layer for AI agents. You bring the harness — the loop that decides what the model does next — and OpenMA takes care of everything else: durable sessions, sandboxed code execution, tool calls, semantic memory, secret-safe credentials, and crash recovery.
Use the SDKs you know: OpenMA supports Claude Managed Agents and OpenAI Agents API. Deploy on any cloud — self-hosted Node + Docker, Cloudflare, or the hosted product at openma.ai. Available capabilities depend on the harness, sandbox adapter, and deployment target.
For the KHA CLI, self-hosted Node also has an experimental, opt-in Kimi Hosted Agents subset (中文指南). It is disabled by default and is not exposed by hosted Cloudflare; check the supported operations and verification limits before migrating.
Most agent frameworks tell you how to build an agent — the loop, the prompt format, the orchestration shape. OpenMA is different: it stays out of the harness and instead provides what every long-running agent needs but is painful to build.
Durable sessions
Append-only event log per session. Resume from anywhere; survive restarts and crashes.
Sandboxed execution
Tool calls run in per-session sandboxes. Pick Docker, E2B, Daytona, Cloudflare Containers, and more.
Vault-backed credentials
Secrets never enter the sandbox. An outbound proxy injects auth headers per-host so tools can hit private APIs without ever seeing the key.
Built-in memory
Per-agent semantic memory across sessions so agents can pick up where they left off.
MCP, everywhere
Connect any MCP server and OpenMA auto-generates mcp_* tools the model can call.
Three integrations, day one
Publish agents into Linear (assignable issues), GitHub (PRs), and Slack (channels) with OAuth-managed auth.