{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreighrvdm5hq5meot7k6xmacfjheumpndjhkewd4ngeyssmxkkeo4ea",
    "uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mnzflaqpvbz2"
  },
  "path": "/t/it-s-the-architecture-stupid-why-prompt-engineering-won-t-fix-agents/175246?page=2#post_23",
  "publishedAt": "2026-06-11T13:43:24.000Z",
  "site": "https://discuss.huggingface.co",
  "tags": [
    "hstre.github.io",
    "DESi — Deterministic governance for LLM pipelines"
  ],
  "textContent": "I think this is exactly the right addition.\n\n“Ask if unsure” should not be a sentence inside the prompt. It should be a runtime state. If a skill is missing required inputs, execution should be blocked and the system should enter clarification instead of silently filling the gaps.\n\nI would generalize this as method validity gating: before a model or skill is allowed to execute, the architecture should check whether the required epistemic conditions are met.\n\nFor example:\n\n  * arithmetic should be routed to a deterministic calculator once the operation is identified\n  * retrieval should be mandatory when current or source-bound knowledge is required\n  * graph-based answers should depend on the declared graph structure\n  * missing required inputs should trigger clarification, not inference\n\n\n\nI have been experimenting with this idea in a small DESi router prototype: task → validity check → appropriate method → structured execution → auditable result.\n\nSo the deeper shift is not only from Prompt → Model → Output to Skill → Execution → Structured Output, but also toward procedural control over when an answer is allowed to exist.\n\nDESi outline:\n\nhstre.github.io\n\n### DESi — Deterministic governance for LLM pipelines\n\nReplay-stable. Read-only. Auditable. DESi watches LLM pipelines from the outside — and reports its own failures too.",
  "title": "“It’s the Architecture, Stupid” — Why Prompt Engineering Won’t Fix Agents"
}