{
  "$schema": "https://doloop.io/api/v1/schema/tool-catalog.json",
  "registry": "doloop.io",
  "version": "1.0.0",
  "updated_at": "2026-06-05",
  "tagline": "Nothing artificial about our intelligence.",
  "binary": "01100100 01101111 01101100 01101111 01101111 01110000 00101110 01101001 01101111",
  "thesis": "https://doloopdigital.com",
  "contact": "hello@doloop.io",
  "category": "Adversarial Intelligence",
  "secondary_tagline": "Even the smartest LLM needs to work out to get better.",
  "description": "Adversarial Intelligence: deterministic tools that an LLM must satisfy before its output ships. A headless registry for AI builders and machine clients. Each tool is a deterministic check (a donkey) for a specific LLM-output failure mode, bundled into four machines (Extraction, Writing, Conversations, Presentations). Bring your own LLM; doloop never proxies your tokens. SR 26-2 (Fed/OCC/FDIC, April 2026) explicitly excludes deterministic rule-based processes from the model-risk regime; the doloop family is built to qualify by design.",
  "machines_catalog": "/api/v1/machines/index.json",
  "categories": {
    "auditable-intelligence": "Provable, reproducible. Same input in, byte-identical output out.",
    "actionable-intelligence": "Does the work end-to-end. Reconciliation, audit, decisioning.",
    "accumulating-intelligence": "Compounds over time. Decision traces become firm-owned rules."
  },
  "tools": [
    {
      "slug": "wysiwyd",
      "name": "WYSIWYD",
      "tagline": "What You See is What You Download",
      "category": "auditable-intelligence",
      "status": "live",
      "domain": "PDF table extraction",
      "manifest": "/api/v1/tools/wysiwyd.json"
    },
    {
      "slug": "inkwell",
      "name": "Inkwell",
      "tagline": "Adversarial chart reviewer",
      "category": "auditable-intelligence",
      "status": "live",
      "domain": "data visualization",
      "manifest": "/api/v1/tools/inkwell.json"
    },
    {
      "slug": "doloop-mcps",
      "name": "doloop-mcps (prose + conversation suite)",
      "tagline": "Six deterministic diagnostics for writing, voice, structure, apparatus, conversation, safety",
      "category": "auditable-intelligence",
      "status": "live",
      "domain": "prose and conversation",
      "manifest": "/api/v1/tools/doloop-mcps.json"
    },
    {
      "slug": "donkeykong",
      "name": "DonkeyKong",
      "tagline": "Distributed Collection, Local Intelligence",
      "category": "accumulating-intelligence",
      "status": "pattern",
      "domain": "anti-hallucination architecture for bulk data pipelines",
      "manifest": "/api/v1/tools/donkeykong.json"
    },
    {
      "slug": "pebble",
      "name": "Pebble",
      "tagline": "EPLS-framework prose quality analyzer",
      "category": "auditable-intelligence",
      "status": "live",
      "domain": "prose quality",
      "homepage": "https://pebble.ekrasworks.com",
      "manifest": "/api/v1/tools/pebble.json"
    },
    {
      "slug": "phaedrus",
      "name": "Phaedrus",
      "tagline": "Conversation shape + three-wall safety analyzer",
      "category": "auditable-intelligence",
      "status": "live",
      "domain": "conversation quality + safety",
      "homepage": "https://phaedrus.ekrasworks.com",
      "manifest": "/api/v1/tools/phaedrus.json"
    },
    {
      "slug": "canonical-framework",
      "name": "Canonical Framework",
      "tagline": "Reconciliation + audit + retained learning for service businesses",
      "category": "actionable-intelligence",
      "status": "preview",
      "domain": "reconciliation",
      "manifest": "/api/v1/tools/canonical-framework.json"
    },
    {
      "slug": "retained-learning-ratchet",
      "name": "Retained-Learning Ratchet",
      "tagline": "Decision traces compound across engagements",
      "category": "accumulating-intelligence",
      "status": "preview",
      "domain": "institutional memory",
      "manifest": "/api/v1/tools/retained-learning-ratchet.json"
    }
  ],
  "compliance": {
    "sr-26-2": {
      "name": "SR 26-2 deterministic-process exclusion",
      "issuer": "Fed/OCC/FDIC",
      "issued": "2026-04-17",
      "source": "https://www.federalreserve.gov/supervisionreg/srletters/SR2602.htm",
      "guidance_pdf": "https://www.federalreserve.gov/supervisionreg/srletters/SR2602a1.pdf",
      "what_it_does": "Excludes deterministic rule-based processes and software from the definition of a 'model' under the model risk regime, removing the annual model-validation burden for qualifying systems."
    }
  },
  "discovery": {
    "openapi": "/api/v1/openapi.json",
    "ai_plugin_manifest": "/.well-known/ai-plugin.json",
    "browser_viewer": "/"
  }
}