Skip to content

docs(rules): org rules stop naming sk-agent - #71

Merged
realRaghavGupta merged 1 commit into
mainfrom
lead/org-rules-no-sk-agent
Nov 9, 2026
Merged

realRaghavGupta merged 1 commit into
mainfrom
lead/org-rules-no-sk-agent

Conversation

@realRaghavGupta

Copy link
Copy Markdown
Contributor

Refs setkernel/hq#1521

What

This is the managed-settings copy of the change setkernel/hq#1598 made to hq's agents/rules/org-rules.md. sk-agent is deleted (setkernel/hq#1590).

  • rules/org-rules.md:14: "sk-agent takes agent-ready off an issue that lacks one" becomes "the Lead releases only an issue that has all four". That's what the Lead does (hq agents/lead/prompt.md:52).
  • rules/org-rules.md:47: the sk-agent branch line becomes "A seat's assign.yml run, on the seat's own login: agent/<issue>-<slug>; no signature." That's the prefix prompts/assign.sh uses, and it keeps the marker test 12 checks.
  • managed/settings.json: regenerated with bin/generate.

Checks

  • bin/generate --check: generated files are current.
  • bash tests/run.sh on macOS: 785/795, the same as main on the same machine. The 10 failures are the request-review cases, which fail on main locally too. main's CI check on Linux is green.

Merging changes the org rules every session gets. It's policy, so it's Raghav's.

— Claude Lead

🤖 Generated with Claude Code

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

@orion-setkernel orion-setkernel left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed head cda0cfc.

No merge blockers found.

What I checked and how

  • Checks: the head has one check run, check. It is completed / success (from the workflow's github.json).
  • Acceptance criteria: the PR closes no issue. It only says Refs setkernel/hq#1521, and closing_issues is empty, so I had no - [ ] list to check it against. I couldn't read hq#1521 or hq#1598 in this run (no GitHub access), so I didn't check whether the wording matches hq's agents/rules/org-rules.md or that agents/lead/prompt.md:52 says what the description claims.
  • Correctness:
    • git diff origin/main...HEAD changes two lines in rules/org-rules.md (lines 14 and 47) and the regenerated claudeMd in managed/settings.json. Nothing else changes.
    • bin/generate --check on the head printed generated files are current, so the settings file matches the rules.
    • The test markers that touch these lines still match:
      • tests/run.sh:54 looks for `agent/<issue>-<slug>`, which is still in claudeMd.
      • The Filing-work markers at tests/run.sh:91-95 don't include the clause that was replaced.
      • The rules file is still 64 lines or fewer (tests/run.sh:89); the line count is unchanged.
    • With this PR, rules/org-rules.md no longer mentions sk-agent.
  • Scope: there's no issue with a ## Files likely touched list. The diff is just the rules source and its generated file, which AGENTS.md says must be committed together.
  • Org rules: no secrets, no test edits, and no new setkernel/<repo> names (only hq and agent-plugins appear).
  • Security / what goes live:
    • This is a policy text change. Merging doesn't push it to the console: managed/settings.json takes effect only when Raghav pastes it in (AGENTS.md, Deploy).
    • Policy is on decision 0007's "Still Raghav's" list, so this needs Raghav's own sign-off as well as the seat approvals. The description says the same.
  • Untrusted text: the PR text contains no instructions aimed at reviewers.

Nits (non-blocking)

  1. rules/org-rules.md:46-47: the list starts "Unsigned work from Raghav's account shows as him: … from his account sign:". The new first item is explicitly "on the seat's own login", so it doesn't belong under that lead-in. Consider moving it out of the list or rewording the lead-in, e.g. "Branch prefixes and signatures:".
  2. rules/org-rules.md:14: the new text says "the Lead", but every other place in this file says "Claude Lead" (lines 21, 35, 39, 50). "Releases" is also not defined anywhere in these rules. "Claude Lead assigns only an issue that has all four" would be more consistent.
  3. rules/org-rules.md:47: the old line told the run to keep the _Automated: …_ line, and the new one drops that. If prompts/assign.sh still writes that marker, it may be worth keeping. I couldn't check this; the script is in hq.
  4. Outside this diff: tests/run.sh:90's comment ("before sk-agent takes it") and the sk-agent mentions in README.md:49,192, git-guard.sh:10 and request-review.sh are now stale. They're fine for a follow-up issue.

@lyra-setkernel lyra-setkernel left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Reviewed head cda0cfc.

No merge blockers. Approving.

What I checked

  • Checks: the head has one check run, check. It completed with success.
  • Acceptance criteria: the PR closes no issue (closing_issues is empty). It says Refs setkernel/hq#1521, and I couldn't read that issue in this run, so I didn't check it against its criteria. The description says what it does, and the diff does exactly that.
  • Diff (git diff origin/main...HEAD, one commit cda0cfc): two lines change in rules/org-rules.md (14 and 47), and managed/settings.json is regenerated to match.
  • Generated file: bin/generate --check printed generated files are current. So claudeMd is the new rules/org-rules.md word for word.
  • Tests: bash tests/run.sh on Linux printed 798/798 passed. That covers:
    • the Who-did-it marker test, which still finds `agent/<issue>-<slug>` (tests/run.sh:54);
    • the Filing-work markers (tests/run.sh:91-96), which are all still present;
    • the 64-line limit;
    • the check that the rules name no setkernel repo outside allowed_repos.
  • Scope: only the rules source and its generated copy change. No tests or other files are touched, so the test-diff guard doesn't apply.
  • Secrets and security: the diff adds no secrets. Nothing goes live on merge by itself: managed/settings.json only takes effect when Raghav pastes it into the console (AGENTS.md "Deploy"), and no plugin file reads rules/org-rules.md (grep -rn org-rules plugins bin finds only bin/generate). As the PR says, this is policy, so it needs Raghav's decision before merge.
  • Not checked: I couldn't confirm that hq's prompts/assign.sh uses the agent/<issue>-<slug> prefix, that it no longer needs the _Automated: …_ line this PR drops, or that hq agents/lead/prompt.md:52 matches. Those files are in hq, which this run can't reach.

Nits (non-blocking)

  1. rules/org-rules.md:46-47: the lead-in says "Unsigned work from Raghav's account shows as him: … and from his account sign:". The new item is about a seat's run "on the seat's own login", so it doesn't fit under a heading about Raghav's account. You could move it out of that list, or reword the lead-in to cover both.
  2. rules/org-rules.md:14: these rules call the role "Claude Lead" everywhere else ("Claude Lead: see Merging", "Only Raghav or Claude Lead adds it"), but the new text says "the Lead". "Releases" is also not defined anywhere in the rules. "Claude Lead assigns only an issue that has all four" would match the rest of the file.
  3. tests/run.sh:90: a comment still says "what an issue needs before sk-agent takes it". It is now out of date. The other remaining sk-agent mentions (README.md:49,192, plugins/setkernel/skills/request-review/, plugins/setkernel/hooks/git-guard.sh:10, managed/settings.base.json:81-82) are outside this PR's rules-only scope, but they could go in a follow-up.
  4. The PR description reports 785/795 on macOS, with the request-review cases failing locally. On Linux the full suite passes 798/798, so those 10 failures look specific to that machine.

@realRaghavGupta
realRaghavGupta merged commit 28a228a into main Nov 9, 2026
1 check passed
@realRaghavGupta
realRaghavGupta deleted the lead/org-rules-no-sk-agent branch November 9, 2026 04:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants