Ship AG2 multi-agent workflows behind APIs with FastAgency
Turn an AG2 multi-agent prototype into a reusable console, web, REST API, or broker-backed service with FastAgency.
npx skills add agentskillexchange/skills --skill ship-ag2-multi-agent-workflows-behind-apis-with-fastagency
pip install fastagency or scaffold a project with pip install cookiecutter and cookiecutter https://github.com/ag2ai/cookiecutter-fastagency.git; define the AG2 workflow, select the UI or adapter, set required LLM API keys, and run the generated tests before deployment.Use FastAgency when an operator has an AG2 multi-agent workflow that should move from notebook or local prototype into a repeatable application surface. The workflow is to scaffold or define a FastAgency project, register the AG2 workflow, choose a user interface such as console or Mesop, expose it through FastAPI or broker-backed adapters when needed, and add tests so the same agent interaction can be exercised in CI before production use. Invoke this instead of using AG2 directly when the work is no longer just agent design and needs a deployable API, app, or operations wrapper around the multi-agent flow. The scope boundary is productionizing AG2 workflows through FastAgency runtime adapters and tests; it is not a generic multi-agent framework, API framework, or AG2 product listing.
What this skill actually does
Inputs and prerequisites: Python 3.10+; FastAgency; AG2/AutoGen workflow; optional Cookiecutter, FastAPI, Mesop, NATS/FastStream, and pytest depending on deployment target.
Setup notes: Install FastAgency with `pip install fastagency` or scaffold a project with `pip install cookiecutter` and `cookiecutter https://github.com/ag2ai/cookiecutter-fastagency.git`; define the AG2 workflow, select the UI or adapter, set required LLM API keys, and run the generated tests before deployment.
Source and verification boundary: use https://fastagency.ai/latest/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Custom Agents workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.