Repository navigation
memory-cleanup: a server it cannot read holds the night open - #411
Merged
Merged
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5d6e9e7e1c
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6
lydakis
pushed a commit
that referenced
this pull request
Oct 30, 2026
lydakis
pushed a commit
that referenced
this pull request
Oct 30, 2026
…st in /background A nightly run that failed while the laptop was offline or a login had expired failed unseen: the trigger recorded "sent" and the turn it sent then failed. Now the window asks the daemon how each trigger's last sent turn ended (one `wait` with timeout 0, every ten seconds, for turns not known yet), and counts as failed: - a trigger whose fire failed, or whose sent turn failed (a turn the person interrupted or steered is not a failure); - a server the nightly memory cleanup could not read, by its `unreadable-SERVER` fact (#411). The footer says `✗ N failed` in red, left of `↻`; the margin glance has a red `✗ N` beside `↻`; both open /background, which lists what failed first with the error under it. The person's own agents' failures (no project in the name, such as memory-cleanup) show in every window. The triggers tool's rows say how the turn their last fire sent ended (`last.turn_status`, `last.turn_error`), so an agent sees what the person sees. No daemon or protocol change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6
lydakis
added a commit
that referenced
this pull request
Oct 30, 2026
…st in /background (#419) * memory-cleanup: a server it cannot read holds the night open Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * FEATURES: #411 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * memory-cleanup: the first night finishes without an unreadable server Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * Sonny: what failed in the background is a red ✗ in the footer and first in /background A nightly run that failed while the laptop was offline or a login had expired failed unseen: the trigger recorded "sent" and the turn it sent then failed. Now the window asks the daemon how each trigger's last sent turn ended (one `wait` with timeout 0, every ten seconds, for turns not known yet), and counts as failed: - a trigger whose fire failed, or whose sent turn failed (a turn the person interrupted or steered is not a failure); - a server the nightly memory cleanup could not read, by its `unreadable-SERVER` fact (#411). The footer says `✗ N failed` in red, left of `↻`; the margin glance has a red `✗ N` beside `↻`; both open /background, which lists what failed first with the error under it. The person's own agents' failures (no project in the name, such as memory-cleanup) show in every window. The triggers tool's rows say how the turn their last fire sent ended (`last.turn_status`, `last.turn_error`), so an agent sees what the person sees. No daemon or protocol change. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * Sonny: a trigger that ended on a failure still shows, and turns are read only for this store's triggers A one-shot that did not deliver, or a trigger stopped by an error, keeps its row when its job goes; /background now lists such rows, scoped by the store id each row now records, so the failure stays until `d` removes it. The triggers tool asks this daemon only about triggers made on its store: a turn handle names a bot and a turn, which another store's daemon has its own of. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * Sonny: /background's pick has a column of its own, apart from the row's mark `›` sat against the row's mark (`›✗`, `›↻`), which read as a glitch. It now takes two columns, as the composer's `› ` does, and every line after a row's first starts under its name. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 * Sonny: an answer that did not reach --reply-to fails in /background, and a start trigger stays its adder's once it ends A turn that went well but whose answer could not be passed on now counts as failed, with the delivery error. A retired `--start` trigger's row keeps the agent that added it (`by`), so its failure stays in that project's windows. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_013jkdGEbQZnXUy1kATtYcQ6 --------- Co-authored-by: Claude <noreply@anthropic.com>
Merged
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Requested by George · project thread
Follow-up to #403, prompted by George's question about what happens when the laptop or a sign-in isn't available.
Most of those cases already catch up:
.saidonly moves when a cleanup finishes, so the next run rereads what was missed.The gap was a chat or mail server failing during the habit step. The cleanup carried on and finished,
since_msmoved past those messages, and nobody ever read them.What changes
The skill text only (memory-cleanup step 5, step 7, and docs/SONNY.md):
referencefact namedunreadable-SERVERwith the error and the time it started failing. It then skipscleanup finish, so the next night starts from the samesince_msand reads what was missed. The open cleanup is the existing cut-short path (duplicate: true).saidpast its 56 KiB window.Not in this PR
Telling the person directly. A fact saved at night reaches a lead only with its next task update, and a night agent has no channel to the person. That needs a design and is raised with George separately.
Checks
Docs and skill only. No code, protocol or schema change.
Generated by Claude Code