Repository navigation
docs(rules): one copy of the org rules for every session and run - #73
realRaghavGupta wants to merge 1 commit into
Conversation
…kernel/hq#1521) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
orion-setkernel
left a comment
There was a problem hiding this comment.
Reviewed head c2fdc18.
What I checked
- Checks: the head has one check run,
check, and it completed withsuccess. No failed or pending checks. - Diff: I read all of
git diff origin/lead/risk-normal-owner...HEAD(one commit, c2fdc18). It touches 5 files:rules/org-rules.md(+12 lines: the merge-blocker review rule, the seats, sessions and Lead-work lines, and a "When a service is down" section), the regeneratedmanaged/settings.json(only theclaudeMdstring changes), the 64→80 line budget intests/run.sh,README.mdandAGENTS.md. - Generated file is in sync:
bin/generate --checkprintsgenerated files are current(rc=0). TheclaudeMddiff is exactly therules/org-rules.mdadditions. No permission or other settings keys change. - Line budget:
wc -l rules/org-rules.mdgives74, which is within the new 80-line limit. The test, README and AGENTS.md all say 80, and no other file in the repo still says 64. - Test edit / org rules: the only test edit raises the line limit in
tests/run.sh. It matches the README/AGENTS.md change and the PR's stated reason. This repo has no test-diff guard (.github/workflows/has onlycheck.yml, and nothing mentionsallow-test-changes), so noallow-test-changes:line is needed. The PR adds no secrets. - Security / what goes live: this is policy. On merge, the managed
claudeMdchanges for every seat. Raghav opened the PR, so it already carries the owner sign-off the rules ask for on an owner path. The rule changes add constraints (how to review, outage handling, who does Lead work) and loosen no permission or deny-list entry.
Not checked
- Acceptance criteria and scope: the PR closes no issue (
closingIssuesReferencesis empty). It onlyRefs setkernel/hq#1521, which I can't read in this run, so I have no- [ ]list or## Files likely touchedto check against. Judged against the PR's own description, the 5 files are what it says it changes. - hq's copy: I couldn't confirm that the added text matches hq's
agents/rules/org-rules.mdor that nothing else there is lost. That repo isn't available in this run. - I didn't rerun
tests/run.shlocally. CIcheckis green on this head.
Blockers
None.
Nits (non-blocking)
rules/org-rules.md:50says the Lead runs as "the timer on raghav-dev". Line 55 gives every raghav-dev sessionraghav-dev/<topic>,— raghav-dev, and line 56 giveslead/,— Claude Leadonly to "Raghav's Mac lead session". So the timer Lead on raghav-dev would now sign asraghav-dev, not asClaude Lead. hq's copy split that out ("On raghav-dev: the Leadlead/<topic>…; Raghav's other sessionsraghav-dev/<topic>"). If that change is intended, fine. If not, one clause on line 55 or 56 would fix it.rules/org-rules.md:20still says "merging (Claude Lead: see Merging)", but the Merging bullet now says only the gate Worker merges and gives the Lead no merge role. This was already there before this PR; a follow-up could drop the parenthetical.
lyra-setkernel
left a comment
There was a problem hiding this comment.
Reviewed head c2fdc18.
What I checked
- Checks: the head has one check run,
check, and it completed withsuccess. That job runstests/run.sh,bin/generate --check, ShellCheck and JSON validation (.github/workflows/check.yml). I also ranbin/generate --checklocally on the head; it printedgenerated files are current, somanaged/settings.jsonmatchesrules/org-rules.md. - Acceptance criteria: the PR only
Refssetkernel/hq#1521 and closes no issue, so there were no- [ ]items to check. I could not read hq#1521 because this run has no GitHub access. Instead I checked the PR against its own description:rules/org-rules.mdgains the four things it says hq's copy alone had: the request-changes rule, the 5 seats with 2–3 sessions per person, Lead work and no pollers, and "When a service is down".wc -l rules/org-rules.mdgives 74, which is at or under the new limit of 80.git grep -e "40–64" -e "64 lines" -e "le 64" HEADfinds nothing, so no old limit is left behind.
- Correctness: the size test in
tests/run.shis still a-letest, now against 80, and README and AGENTS.md say the same number. The generatedclaudeMdmatches the source file. - Scope: five files changed: the rules, the generated settings, the line limit in README, AGENTS.md and
tests/run.sh. That is what the PR describes. Nothing else changed. - Org rules and security: the diff has no secrets. This repo has no test-diff guard:
check.ymlis the only workflow, and nothing under.github/orbin/mentionsallow-test-changes. Thetests/run.shedit only moves the limit, as the PR says. What goes live on merge is the managedclaudeMd, which reaches every seat. That is policy and an owner path. The PR says Raghav merges it, and he opened it.
Blockers
None.
Nits (non-blocking)
-
Which signature the raghav-dev Lead uses. The new line says the Lead is "the timer on raghav-dev, or a session Raghav starts as Lead". Further down, the signing list only gives
lead/and— Claude Leadto "Raghav's Mac lead session". So once hq's copy is deleted, the timer Lead on raghav-dev will match the hostname rule and signraghav-dev/and— raghav-dev, the same as Raghav's other sessions there. Before this change, hq's copy gave itlead/. If that is what you intend, fine. If not, add one line saying how the raghav-dev Lead signs, so its work can be told apart from his other sessions. -
Details from hq's copy that this file does not keep:
- the seats are "the members of the GitHub team setkernel/seats";
- setkernel-ops[bot] sets board Needs and copies ticked criteria from the PR body;
- Lead work includes "(re)assigning" issues.
The PR says this file's wording wins where the copies differ. If any of these is still how things work, it is now written nowhere once hq deletes its copy.
Not checked
- The text of setkernel/hq#1521 and of hq's
agents/rules/org-rules.md: there was no GitHub access in this run. I compared against the hq rules text this session was given. - Whether the hq runner PR really fetches this file: that is a separate PR.
- I did not run the full
tests/run.shsuite. I relied on the green CI run on this head.
Refs setkernel/hq#1521 (F7). Stacked on #72: its base is #72's branch, so review this diff alone and merge it after #72. GitHub retargets it to main when #72 merges.
Why
Raghav decided on 2026-11-09 that this file is the one source of the org rules. hq's copy,
agents/rules/org-rules.md, is the one every headless run (the Lead on raghav-dev, seat assign.yml runs) gets today. The managed settings don't reach those runs: vega-dev'sremote-settings.jsonis empty. A setkernel/hq PR makes both runners fetch this file and deletes hq's copy.What
This adds the rules only hq's copy had, so nothing is lost when it goes:
Where the copies differed, this file's wording and branch prefixes stay (the Mac Lead session is
lead//— Claude Lead).Budget: the file is 74 lines, so the limit in
tests/run.sh, README and AGENTS.md goes from 64 to 80. That's the size hq's copy (80 lines) already had in every run.Checks
bin/generate --check: current.bash tests/run.shon macOS: 776/786. The 10 failures are the request-review cases that fail on main on this machine too.Policy, so it's Raghav's merge.
— Claude Lead
🤖 Generated with Claude Code