{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreih2maenyk324k7nsemhb3sjmuuynsawf7bb74qejptrhznkm7puz4",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mmjls7x7bpq2"
  },
  "path": "/t/what-if-openai-agent-builder-had-reusable-cognition-orca-mcp-demo/176178#post_1",
  "publishedAt": "2026-05-23T13:39:49.000Z",
  "site": "https://discuss.huggingface.co",
  "tags": [
    "https://github.com/gfernandf/agent-skills",
    "https://zenodo.org/records/19438943",
    "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6600840"
  ],
  "textContent": "What if **OpenAI Agent Builder** could do something it doesn’t natively support yet?\n\n**Reusable, auditable cognition.**\n\nI built a small ORCA-powered agent on top of **OpenAI Agent Builder + MCP** that can take a strategic decision request like:\n\n> _“Should we launch a legal SaaS in Spain during the next 12 months?”_\n\nand instead of producing a black-box answer, it executes an explicit cognitive workflow:\n\n**prompt → routing → reusable decision skill → execution trace → confidence → report**\n\nThe interesting part is not the recommendation.\n\nThe interesting part is that the reasoning becomes:\n\n**reusable**\n**traceable**\n**auditable**\n**composable**\n\nAnd the output stops looking like:\n\n> _“here are some thoughts…”_\n\nand starts looking closer to something a real team could actually use:\n\n  * **explicit recommendation**\n\n  * **alternatives evaluated and scored**\n\n  * **confidence levels**\n\n  * **decision quality assessment**\n\n  * **uncertainties and missing information**\n\n  * **execution diagnostics**\n\n  * **cognitive trace of what happened**\n\n  * **graceful degradation when execution fails**\n\n\n\n\nUnder the hood, the agent delegates cognition to an **ORCA skill exposed through MCP**.\n\nThe surprising thing?\n\nIt is actually much more practical than I expected to expose a reusable skill via MCP and plug it into Agent Builder.\n\nBut it already feels surprisingly close to something agent platforms will eventually need.\n\nBecause once agents become business-critical:\n\n> **“trust me bro, the model reasoned”** probably won’t be enough.\n\nThis is part of **ORCA (Open Cognitive Runtime for Agents)** — an open-source framework in progress for reusable cognition in agents.\n\nRepo:\nhttps://github.com/gfernandf/agent-skills\n\nPaper:\nhttps://zenodo.org/records/19438943\n\nhttps://papers.ssrn.com/sol3/papers.cfm?abstract_id=6600840\n\nCurious question for people building agents:\n\n**Should cognition stay inside prompts, or should it become a reusable runtime layer?**",
  "title": "What if OpenAI Agent Builder had reusable cognition? (ORCA + MCP demo)"
}