Back to docs home
Examples Service docs
Synced from the macp-playground repo. Covers scenario authoring, launch compilation, agent hosting, and the worker bootstrap contract.
All documents
10 documents.
- Adding a Framework HostThis guide explains how to add support for a new agent framework (e.g., AutoGen, Semantic Kernel, OpenAI Agents SDK).
- API ReferenceAll endpoints return JSON. Error responses follow the format:
- ArchitectureThis page covers only the macp-playground internals (catalog, compiler, hosting). For the MACP runtime's own architecture — layer structure, request flow, durability model, and mode/policy registries — see the canonical doc: [`macp-runtime/docs/architecture.md`](https://github.com/multiagentcoordinationprotocol/macp-runtime/blob/main/docs/architecture.md). For protocol-level concepts (sessions, modes, two-plane model) see the [protocol docs](https://www.multiagentcoordinationprotocol.io/docs).
- DeploymentThe repo is **fully platform-agnostic**. The `Dockerfile` is the only deployment contract — no platform-specific config files in the repo. All platform settings live in each platform's dashboard.
- Direct-agent-auth in the macp-playgroundThis document describes how the **macp-playground** spawns agents under [RFC-MACP-0004 §3](https://github.com/multiagentcoordinationprotocol/multiagentcoordinationprotocol/blob/main/rfcs/RFC-MACP-0004-security.md#3-authentication) ("the `sender` field MUST be derived from authenticated identity"): how scenarios compile, how per-agent JWTs are minted, how runs are registered with the control-plane, and how ambient envelopes are authorized. It is the canonical home for AUTH-2 minting, CP-1 behaviour and ambient-envelope scopes; policy registration is documented in [`policy-authoring.md`](policy-authoring.md#how-policies-are-registered) and the bootstrap file shape in [`worker-bootstrap-contract.md`](worker-bootstrap-contract.md).
- Framework Hosting DesignThe MACP Playground uses a **framework-neutral hosting architecture** that allows heterogeneous agent frameworks (LangGraph, LangChain, CrewAI, custom) to participate in the same MACP run through a unified control-plane contract.
- Policy Authoring GuidePolicies define the governance rules that control how specialist agent signals (evaluations and objections) are aggregated into final decisions during MACP coordination runs. This guide explains how the **macp-playground** loads policies, how they flow to spawned agents, and how to author new ones for demo scenarios.
- Scenario Authoring CLIA small developer-facing CLI that lets you scaffold, validate, dry-run, and lint scenario packs without booting the HTTP server. Internal authoring tool — not exposed over HTTP.
- Scenario Authoring GuideEach scenario pack follows this layout:
- Worker Bootstrap ContractEvery MACP worker process receives a **bootstrap payload** — a JSON file containing everything needed to start participating in a run. The shape is flat and framework-agnostic; both `macp_sdk` (Python) and `macp-sdk-typescript` consume it verbatim via their `fromBootstrap()` / `from_bootstrap()` entry points.