Repository navigation
fix(ops): sample deeper when skipped guard runs hide the deploy verdict - #118
Merged
Merged
Conversation
The 2026-09-02 check-in archive reported `Deploy to Cloudflare` as `missing` while the deploy for that SHA had succeeded. The classifier was right and the sample was too shallow. `daily-checkin-data.mjs` asked `gh run list --workflow <name> --limit 10`. A `workflow_run` event from every branch triggers the deploy workflow, so its guard writes a `skipped` run on every push. On 2026-09-01 eleven consecutive skipped runs sat in front of the success, one row past the page edge, so the sample carried no verdict at all. GitHub cannot answer "the last run that was not skipped". Both `gh run list --status` and the REST `?status=` filter take one value, while a verdict can be success, failure, cancelled, timed_out, neutral, action_required or stale. The exclusion has to be a client-side read of ordered history. `collectWorkflowRuns` keeps the 10-run head page and asks once more for 100 runs only when that page was full and reached no verdict. A short head page is the workflow's whole history, so it is never re-fetched. 100 is one GitHub page, so a workflow with no verdict in 100 runs has genuinely not reported one and `missing` stays meaningful. Claude-Session: https://claude.ai/code/session_013pSLxovqDjiP7U7rUrRh9n
harlan-github-agent
Bot
added
harlan-agent-running
Contributor
🤖 MERGED
GitHub merged this pull request. No further automated Review will run. |
harlan-github-agent
Bot
added
harlan-agent-ready
harlan-github-agent
Bot
added
harlan-agent-running
harlan-github-agent
Bot
removed
the
harlan-agent-pending
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
1 participant
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.
❓ Type of change
📚 Description
The 2026-09-02 check-in archived
Deploy to Cloudflareasmissingwhile the deploy for that SHA had already succeeded and production was serving it.missingis meant to say the check-in could not see a verdict, so firing it on a healthy deploy teaches the reader to skip the line.The sample was the problem, not the classifier. A
workflow_runevent from every branch triggers the deploy workflow and its guard skips itself offmain, so the 10 runs the collector asked for on 2026-09-01 were all guard runs:It now asks for 100 runs, but only for a workflow whose first page came back full and carried no verdict, so the ordinary case still costs one request each.
I wanted to ask GitHub for the last non-skipped run rather than pick a number. It cannot answer that:
gh run list --statusand the REST?status=filter both take a single value, and a verdict can be any of seven conclusions. The exclusion stays client side and both numbers are commented where they live.Open question I did not resolve: 100 is one GitHub page and it keeps
missinghonest, but I have no measurement of how long a skipped streak gets on a busy day. If one ever passes 100, this comes back asmissingjust as quietly.Ledger row added under
Deploy verdict falls off the end of the run sample.https://claude.ai/code/session_013pSLxovqDjiP7U7rUrRh9n