# OpenWOP > OpenWOP (Workflow Orchestration Protocol) is an open-source protocol for multi-agent workflow orchestration. It defines how a run of AI agents, tools and people starts, streams its events, stops for a person's approval, resumes and replays, over plain REST and Server-Sent Events, so it behaves the same on any host that implements it. The specification, schemas and API definitions are CC BY 4.0; the SDKs (TypeScript, Python, Go), conformance suite, CLI and packs are Apache 2.0. Every page listed here also has a markdown copy at the same URL with `.md` appended (`index.html.md` for a URL ending in `/`). Use OpenWOP when you build AI agent workflows that take real actions (payments, messages, record changes) and need human-in-the-loop approval, an audit trail, replay of past runs, and freedom to switch hosts. It is a protocol, not a framework: any host that implements it offers the same run lifecycle and event log. It complements MCP (agent-to-tool) and A2A (agent-to-agent); an OpenWOP host can call MCP tools and A2A agents from inside a run. ## Start here - [Quickstart](https://openwop.dev/quickstart/): start the v2 reference host locally, create a run over HTTP, stream its events, then approvals, replay and webhooks - [Hosted app](https://app.openwop.dev/): the reference host in production; try it with no sign-up - [OpenWOP for AI agents](https://openwop.dev/ai-tools/): skills an agent can follow to discover a host, start a run, handle an approval, fork a run and check conformance, plus safety rules - [The protocol in one page](https://openwop.dev/protocol/): every protocol surface, linked to its spec - [Comparison with A2A and MCP](https://openwop.dev/comparisons/a2a-openwop-mcp/): which protocol does which job - [Alternatives](https://openwop.dev/comparisons/): fair comparisons with LangGraph, CrewAI, AutoGen / Agent Framework / AG2, Temporal and n8n, each saying where the other tool is the better choice ## Specification - [v2 specification](https://openwop.dev/spec/v2/): core documents every host implements, plus optional extensions - [Runs](https://openwop.dev/spec/v2/core/runs.html): create, read, stream, cancel, pause, fork and diff a run - [Events](https://openwop.dev/spec/v2/core/events.html): the append-only run event log and its envelope - [Human approval (interrupts)](https://openwop.dev/spec/v2/core/interrupt.html): suspend a run for a person and resume it - [Replay and fork](https://openwop.dev/spec/v2/core/replay.html): re-run from any past step with side effects suppressed - [Capabilities](https://openwop.dev/spec/v2/core/capabilities.html): how a host advertises what it supports at /.well-known/openwop - [Errors](https://openwop.dev/spec/v2/core/errors.html): the error-code registry - [REST API reference](https://openwop.dev/api/rest/): OpenAPI for the v2 surface - [Event API reference](https://openwop.dev/api/events/): AsyncAPI for streamed events ## Build on it - [For workflow authors](https://openwop.dev/for/workflow-authors/): call any compatible host over REST and SSE - [For host implementers](https://openwop.dev/for/host-implementers/): implement the core and prove it with the conformance suite - [For pack authors](https://openwop.dev/for/pack-authors/): publish signed, reusable steps and agents - [CLI](https://openwop.dev/cli/): drive any host from a terminal - [Tool calling](https://openwop.dev/tool-calling/): effect identity, retries and at-most-once side effects ## Evidence and project - [Conformance leaderboard](https://openwop.dev/conformance/): which hosts implement it and the signed evidence behind each claim - [RFCs](https://openwop.dev/rfcs/): every proposal that changed the protocol, by status - [Versioning](https://openwop.dev/versioning/): additive within a major; breaking changes only in a new major - [Governance](https://openwop.dev/governance/): how decisions are made - [Source on GitHub](https://github.com/openwop/openwop): specification, schemas and conformance suite ## Optional - [Full text for LLMs](https://openwop.dev/llms-full.txt): the v2 core and extension specs plus the site's guides in one markdown file - [FAQ](https://openwop.dev/faq/) - [Scenario walkthroughs](https://openwop.dev/scenarios/) - [Security posture](https://openwop.dev/security/) - [Roadmap](https://openwop.dev/roadmap/)