Back to Spatial Atlas

Card snapshot

arun0808-spatial-atlas.hf.space · 2026-05-18 15:45:56 UTC · dec01ce0c94768762f162b4d25a4a78e67e4a574e0da25d1258f01180c3f8e3d

This is a frozen copy of the agent's agent-card.json as we observed it at the timestamp above. We capture a new snapshot every time the card's content hash changes. Useful for: forensic drift analysis, verifying downstream callers see the right version, reproducing routing decisions made historically.

{
  "capabilities": {
    "streaming": true
  },
  "defaultInputModes": [
    "text"
  ],
  "defaultOutputModes": [
    "text"
  ],
  "description": "Spatial-aware research agent built on compute-grounded reasoning (CGR). Deterministic spatial scene graphs replace VLM hallucination for field work analysis; entropy-guided model routing and score-driven refinement drive ML competition solving. A2A-compliant for AgentBeats Phase 2 Sprint 2.",
  "name": "Spatial Atlas",
  "preferredTransport": "JSONRPC",
  "protocolVersion": "0.3.0",
  "skills": [
    {
      "description": "Analyzes factory, warehouse, and retail environments from images, videos, PDFs, and documents. Spatial reasoning with structured scene graphs, safety inspection, and formatted reporting.",
      "examples": [
        "Analyze warehouse layout for safety violations"
      ],
      "id": "fieldwork-research",
      "name": "Multimodal Field Research",
      "tags": [
        "spatial",
        "multimodal",
        "vision",
        "fieldwork",
        "research"
      ]
    },
    {
      "description": "Solves Kaggle-style ML competitions end-to-end: data analysis, feature engineering, model training, and submission generation.",
      "examples": [
        "Train a model for the spaceship-titanic competition"
      ],
      "id": "ml-engineering",
      "name": "ML Engineering",
      "tags": [
        "ml",
        "kaggle",
        "data-science",
        "code-generation"
      ]
    }
  ],
  "url": "https://arun0808-spatial-atlas.hf.space/",
  "version": "1.0.0"
}