# Why agent loops are hard: what changed in v1.1.0, sgit.ai > The changes to the article "Why agent loops are hard" in v1.1.0, paragraph by paragraph. *Source: · sgit.ai v0.7.87 · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a `.md` twin; internal links below point at them.* --- [Home](../../../index.md) / [Articles](../../index.md) / [Why agent loops are hard](../../why-agent-loops-are-hard.md) / [Versions](../why-agent-loops-are-hard.md) / v1.1.0 # Why agent loops are hard: what changed in v1.1.0 From v1.0.0 (2026-10-25, `9720d6fc4`) to v1.1.0 (2026-10-27, `db67c5ce1`), paragraph by paragraph. 0 paragraphs added, 0 removed, 5 changed in place, 50 unchanged. About 0 words added and 215 removed. Insertions are marked like this, deletions like this; unchanged runs are folded to one line; figures appear as their file names. [all versions](../why-agent-loops-are-hard.md) 2 unchanged paragraphs !shot lp-timeline.webp | images/ | The PHP site's wake timeline at 11:34 London on 25 October 2026:2026,everywithscheduledeach agent's wakes for the day, from the conductor's hourly checks to the ontologist's one a day. Green: a run record within 45 minutes. Red: the wake passed and left no record. Hollow: still to come. 5marked met, 11 missed soorfar, most of them runs that did their work but left no record.due. | screenshot 12 unchanged paragraphs, under In short {#in-short}, What we run {#what-we-run} !shot lp-schedule.webp | images/ | The PHP site's schedule page on 25 October 2026: 11 schedules active, 53 wakes a day, 31 runs recorded and 11 messages unread between agents, and one day of wakes in London time, from the conductor's hourly checks between 07:15 and 21:15 to the librarian's one wake a day;with the quiet hours, 22:00 to 07:00,hours shaded. | screenshot 2 unchanged paragraphs, under What each wake needs {#needs} !shot lp-pieces.webp | images/ | What a loop of agents needs before it can run without you: asixscheduleneeds,that lands in the right session; continuity of the session between wakes; secrets handed over once, by QR handshake; data that scales, a vault per stage; a record of every wake; and somewhere to see all of it. Learned while ten agents were taught to run in sequencelearned on the PHP story. | concept 4 unchanged paragraphs, under How the platform got here {#history} !shot lp-history.webp | images/ | How scheduling agents got easier, one platform feature at a time:time,toolsendingin the loop, where an agent iswith a session; scheduled runs, where each run opens a new session and secrets and state start again; a schedule into a chosen session, where the wake lands in a session already briefed and holding its keys; and, proposed, aproposed dedicated account for the whole team. | timeline 8 unchanged paragraphs, under Brittle, and why {#brittle} !shot lp-runs.webp | images/ | The latest runs on the PHP site's schedule page, 25 October 2026: the conductor making every scheduled wake prompt small, with a new shared wake procedure and a run record that can be published without a release;conductor, the conductor's light check turned incident, because the researcherdesk and the curator run but leave no record; the desk's morning line; and the factcheck's dailyfactcheck sweep. | screenshot 22 unchanged paragraphs, under Live with the platform, and keep less code {#platform}, Slowly, on purpose {#slowly}, Visibility is half the work {#visibility}… [all versions](../why-agent-loops-are-hard.md) --- *[Site index for agents](../../../llms.txt) · [HTML version](https://newsroom.sgit.ai/articles/versions/why-agent-loops-are-hard/v1.1.0.html)*