← Public packages
@kentcdodds/closing-the-agentic-loop
Closing the Agentic Loop: Kent C. Dodds's talk and workshop deck. Default short talk; ?event=aie-2026 is the AIE hour cut with a Podium QR; ?format=full adds labs.
Files
{
"note": "Hour slide set (format: hour). Applied when an event uses this format, or via the ?format=hour alias. deck.js hides HOUR_HIDE, adds data-format=hour slides, then applies these overrides: page-title/event fallbacks are on the event; these notes and moveBefore are the hour cut. Each notes entry replaces the one speaker-notes paragraph (on that slide) whose text starts with `starts`; `text: null` removes it.",
"moveBefore": {
"your-turn": "final"
},
"notes": {
"title": [
{
"starts": "Open: \"Light slides today.",
"text": "Open: one hour today. I explain what autopilot does instead of running it, and trying it on your own repo is your exercise."
}
],
"factory": [
{
"starts": "Then the line on screen:",
"text": "Then the line on screen: three pillars make this safe, not just fast. That's what this session is about."
},
{
"starts": "Transition:",
"text": "Transition: \"Here's the loop that makes that possible.\" Next slide: the thesis."
}
],
"pillars": [
{
"starts": "Next slide: start the live autopilot run",
"text": "Next: pillar 1. No autopilot demo today; one slide after the ladder explains what it does, and trying it is the exercise."
}
],
"section-1": [
{
"starts": "Pillar 1. Next slide: start the live autopilot run",
"text": "Pillar 1. Next: the ladder, then one slide on what autopilot does. I explain it rather than run it; running it on your own repo is the exercise."
},
{
"starts": "Timing: a 3-minute Q&A",
"text": null
}
],
"ladder": [
{
"starts": "Leave this up while the audit runs.",
"text": "Order matters: P1 makes everything checkable; P2 and P3 earn trust for agent merges; the Keep improving rungs amplify what exists. Next slide: what autopilot does with these rungs."
},
{
"starts": "mediarss read-only dry run",
"text": "mediarss read-only dry run (Oct 18, run with kcd-skills a014c09): done 0 \u00b7 partial 9 \u00b7 missing 3 \u00b7 unknown 0 across 12 rungs (missing: Decisions directory, Cleanup issues, Cleanup cron agents; rungs 2, 12 and 13 on this ladder). The dry run predates the Triage rung (11), so it has no score for it. Unknown means \"ask or verify\"; it never becomes a PR until resolved."
}
],
"primitives": [
{
"starts": "Timing: about 2 minutes while the audit runs",
"text": "Timing: about 2 minutes."
}
],
"section-2": [
{
"starts": "Pillar 2:",
"text": "Pillar 2: ship-pr, one real agent PR (#3782), then how agents see the app, and health checks (health --sha after deploy)."
}
],
"pr-3782": [
{
"starts": "No screenshot in this PR.",
"text": "No screenshot in this PR. Next: how an agent proves its own change."
}
],
"section-3": [
{
"starts": "Pillar 3",
"text": "Pillar 3: one slide on flags, tested restores and PocketOS (rungs 8 and 9). Health checks and the deploy guards were covered at the end of pillar 2."
}
],
"section-4": [
{
"starts": "Loops and the factory:",
"text": "Loops and the factory: the friction loop, the signal loop, and how a product decision happens. They keep the factory from the opener running."
},
{
"starts": "Then back to the run",
"text": "Then Your turn (the exercise), the final slide, and Q&A."
}
],
"your-turn": [
{
"starts": "Short version only.",
"text": "Hour mode: this slide moves to right before the final slide. I explained autopilot; running it on your own repo is the exercise."
},
{
"starts": "Optional, for anyone with a laptop.",
"text": "Optional, for anyone with a laptop, now or after the session. Leave it up for a moment, then the final slide and Q&A."
}
],
"product-decision": [
{
"starts": "For a live run, pick a card",
"text": "Hour mode: show the offline capture of the voice walkthrough rather than a live demo."
}
]
}
}