skills.rest
sandgardenhq
sandgardenhqOFFICIAL
Explore all skills by this author →
GitHub

using-sgai

Automate sgai software factory workspaces via probe, poll, and act loops.

130|8|Updated Jan 23, 2026
Search Tags:#workflow orchestration#mcp#agent coordination#human-in-the-loop#sse#http api#sgai
One-click install
npx skills add https://github.com/sandgardenhq/sgai --skill using-sgai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: using-sgai
Source: https://github.com/sandgardenhq/sgai/tree/main/docs/sgai-skills/using-sgai
Command: npx skills add https://github.com/sandgardenhq/sgai --skill using-sgai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coordinating an sgai software factory from an external AI harness can be error-prone without a clear control loop. This Skill provides a concise, repeatable probe/poll/act pattern to start sessions, monitor progress, and handle human-in-the-loop questions across workspaces.

Core Features & Use Cases

  • Probe/Poll/Act Loop: Continuously fetch factory state, detect blocking questions, and take the right action for each workspace.
  • Human Interaction Handling: Respond to free-text, multi-choice, and work-gate questions to unblock agents.
  • HTTP and MCP Parity: Drive sgai via REST endpoints or the MCP tool interface, including real-time updates via SSE.
  • End-to-End Runbook: Create a workspace, set a goal, start in auto mode, and monitor to completion.

Quick Start

Ask the assistant to connect to your running sgai server, list workspace state at /api/v1/state, and respond to any pendingQuestion using the appropriate respond endpoint or MCP tool.

Frequently Asked Questions about using-sgai

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate multi-agent software development sessions from an external IDE?▼

You can orchestrate multi-agent software development sessions by driving an sgai software factory via HTTP API or MCP tool interface, using a probe/poll/act loop to start sessions, monitor progress, and handle human-in-the-loop questions.

What is the best way to handle human-in-the-loop questions during agent workflow orchestration?▼

Handling human-in-the-loop questions during agent workflow orchestration involves polling workspace state, detecting pending questions, and responding to free-text, multi-choice, or work-gate prompts via the appropriate respond endpoint or MCP tool to unblock agents.

Can I monitor sgai workspace state using SSE instead of periodic polling?▼

Yes, you can monitor sgai workspace state using real-time Server-Sent Events (SSE) for updates, or you can periodically poll the /api/v1/state endpoint to fetch factory state and detect blocking questions.

Do I need a running sgai server to use this MCP agent coordination skill?▼

Yes, using this MCP agent coordination skill requires a running sgai server with HTTP API access at /api/v1 and an optional MCP endpoint at /mcp/external to manage workspaces and invoke start or respond actions.

How does the probe poll act loop work for coordinating AI agents over HTTP APIs?▼

The probe/poll/act loop for coordinating AI agents continuously fetches factory state via HTTP APIs, detects blocking questions, and takes the appropriate action for each workspace to automate control of multi-agent software development sessions.

Does sgai support both REST endpoints and MCP tools for workspace automation?▼

Yes, sgai supports HTTP and MCP parity for workspace automation, allowing you to drive software factory operations via REST endpoints or the MCP tool interface to start sessions, monitor progress, and respond to human-in-the-loop questions.

Related Skills

Explore More Software Engineering Skills→
sandgardenhqsandgardenhq
130

monitoring

Monitor sgai workspaces and agent activity via HTTP and SSE endpoints.

Official
Basic
RHEcosystemAppEngRHEcosystemAppEng
48

debug-pipeline

Diagnose OpenShift Pipelines (Tekton) failures by analyzing PipelineRun status and TaskRun logs.

Official
Advanced
ldmrepoldmrepo
3

agui

Build real-time streaming AI agent applications using the AG-UI protocol.

Community
Intermediate
ucirelloucirello
2

using-sgai

Manage Sgai workspaces and sessions via HTTP API and MCP interface.

Community
Advanced