Build bounded single-binary agent workflows with Yao
Use Yao when an operator needs to package an AI agent workflow, hooks, MCP tools, memory scopes, and a reviewable web or chat interface into one deployable runtime.
npx skills add agentskillexchange/skills --skill build-bounded-single-binary-agent-workflows-with-yao
Invoke this workflow when a team needs a controlled agent runtime rather than another loose prompt or ad hoc service. The operator defines Yao resources, chooses the executor mode, wires TypeScript Create and Next hooks, connects MCP tools or CLI agents, and exposes the result through the built-in chat UI, REST API, or app surface. Reviewable inputs include the agent goal, hook code, model provider, memory scope, tool definitions, and deployment target. Reviewable outputs include the packaged single-binary service, route behavior, hook logs, agent responses, and any downstream workflow actions.
What this skill actually does
Use this instead of the product normally when an agent needs a narrow operational boundary: route requests before execution, validate outputs after execution, isolate CLI agents in a sandbox, or mix deterministic hooks with LLM and tool calls. The scope boundary is the repeatable build-and-operate workflow for Yao agent runtimes. It is not a broad low-code platform entry; the skill is specifically about defining, running, and reviewing bounded agent services with Yao’s documented hooks, memory API, MCP support, and single-binary deployment model.