What problem does it solve? Starting work in an SGAI-managed repository requires a correctly structured GOAL.md with an agent flow graph, model assignments, a completion gate, and outcome-based success criteria, which is error-prone to write by hand. ## Core Features & Use Cases - Stack Detection: Scans for go.mod, package.json, wrangler.toml, and similar files to select the right developer and reviewer agents. - Flow Graph Composition: Builds developer-to-reviewer-to-stpa-analyst agent chains, including multi-stack variants, backed by a full agent catalog reference. - Safe File Handling: Saves model preferences to .sgai/preferences.json and backs up any existing GOAL.md with a timestamped name before writing. - Use Case: You want to add OAuth authentication to a Go plus React app. The skill detects both stacks, drafts outcome-based success criteria, assembles a four-agent flow graph ending in stpa-analyst, sets go test as the completion gate, and writes the GOAL.md after your approval. ## Quick Start Ask the agent to create an SGAI goal for adding user authentication to this repository.