{
  "schemaVersion": 1,
  "name": "Agent Workflow Canvas",
  "author": {
    "name": "Kevin Astuhuaman",
    "portfolio": "https://portfolio.kevinastuhuaman.com",
    "github": "https://github.com/kevinastuhuaman"
  },
  "thesis": "Enterprise AI should make uncertainty, authority, and recovery visible before an agent takes consequential action.",
  "users": [
    "operations teams",
    "product managers",
    "non-technical workflow owners"
  ],
  "productDecisions": [
    "Show the workflow and execution trace in the same product surface.",
    "Treat human approval as a stateful policy boundary, not a confirmation modal.",
    "Invalidate progress after a permission failure and make recovery explicit.",
    "Use deterministic synthetic scenarios so every state is inspectable."
  ],
  "capabilities": [
    "agent workflow design",
    "human-in-the-loop approval",
    "execution traces",
    "confidence routing",
    "fail-closed permissions",
    "recovery",
    "enterprise UX"
  ],
  "scenarios": [
    {
      "id": "mismatch",
      "outcome": "paused for accountable finance approval"
    },
    {
      "id": "clean",
      "outcome": "completed through the automatic branch"
    },
    {
      "id": "permission",
      "outcome": "failed closed, then recovered with scoped access"
    }
  ],
  "evidence": {
    "liveProduct": "https://kevinastuhuaman.github.io/agent-workflow-canvas/",
    "source": "https://github.com/kevinastuhuaman/agent-workflow-canvas",
    "decisions": "https://github.com/kevinastuhuaman/agent-workflow-canvas/tree/main/DECISIONS",
    "tests": "https://github.com/kevinastuhuaman/agent-workflow-canvas/tree/main/tests"
  },
  "publicationMode": "open prototype",
  "data": "synthetic",
  "ipBoundary": "Original demonstration code only; no employer code, customer data, production credentials, private prompts, or proprietary automation logic."
}