docs(agents): render the managed AGENTS.md git-workflow block

coilyco-ops commented

2026-08-26 02:48:17 +00:00

Owner

The hand-written one-line git-workflow stamp drifted from the canonical renderer in
agentic-os, and on a pull-request lane it inverted who merges: the agent that
authored the code merges its own pull request, and reading the lane the other way
leaves finished work sitting unmerged.

Rendered from the lane this file already declares in its own frontmatter, so nothing
per-repo chose it. The applier strips the legacy stamp on its own.

One-time rollout from coilyco-flight-deck/infrastructure#937

Lane: pull-request-and-merge. Not merged by this run.

The hand-written one-line git-workflow stamp drifted from the canonical renderer in agentic-os, and on a pull-request lane it inverted who merges: the agent that authored the code merges its own pull request, and reading the lane the other way leaves finished work sitting unmerged. Rendered from the lane this file already declares in its own frontmatter, so nothing per-repo chose it. The applier strips the legacy stamp on its own. One-time rollout from https://forgejo.coilysiren.me/coilyco-flight-deck/infrastructure/issues/937 Lane: `pull-request-and-merge`. Not merged by this run.

coilyco-ops added 1 commit

2026-08-26 02:48:17 +00:00

docs(agents): render the managed AGENTS.md git-workflow block

ci / test (pull_request) Failing after 1m16s

Details

ci / publish-echo-image (pull_request) Has been skipped

Details

ci / publish-observed (pull_request) Has been skipped

Details

ci / image-build (pull_request) Failing after 49s

Details

ea6326ba9a

The hand-written one-line git-workflow stamp drifted from the canonical renderer in
agentic-os, and on a pull-request lane it inverted who merges: the agent that
authored the code merges its own pull request, and reading the lane the other way
leaves finished work sitting unmerged.

Rendered from the lane this file already declares in its own frontmatter, so nothing
per-repo chose it. The applier strips the legacy stamp on its own.

One-time rollout from coilyco-flight-deck/infrastructure#937

Co-authored-by: Kai Siren <coilysiren@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
Agent-Role: platform

coilyco-ops commented

2026-08-26 02:55:22 +00:00

Author

Owner

Blocked, and not by the change in this pull request. Leaving it open with the evidence rather than merging red or closing it.

What CI says

ci / test fails in role-drift-check:

a role's selection changed: agent/rendered/roles/devrel.bundle.txt,
  agent/rendered/roles/eval.bundle.txt, agent/rendered/roles/frontend.bundle.txt,
  agent/rendered/roles/gamedev.bundle.txt, agent/rendered/roles/platform.bundle.txt,
  agent/rendered/roles/sysadmin.bundle.txt, agent/rendered/roles/tpm.bundle.txt

Every name in that list is a role this repo has never recorded.

Why it is not this change

The recorded roster on main is the old naming:

ai creator design director engineer exec librarian ops qa

A bake against the current catalog produces the Core Roster v2 seats instead: devrel, eval, frontend, gamedev, platform, sysadmin, tpm. That is an upstream rename, and nothing in this branch touches roles. The branch changes AGENTS.md and nothing else.

scripts/task.sh clones the catalog at ${AOS_CATALOG_REF:-main}, unpinned, so the drift check compares this repo's committed record against whatever agentic-os main holds at the moment CI runs. agentic-os main took the seat retitles, so the record went stale without a commit here.

main is green only because it has not re-run since. Any pull request opened in this repo right now fails the same way, this one included.

What it needs

A rebake and re-record of the new roster, which is the seat-rename migration rather than a drive-by on a docs change. I did not do it: the bake is only as correct as the catalog it reads, and recording a roster from an arbitrary local checkout would bake unrelated upstream drift into this repo's committed record.

Filed the unpinned-ref half separately. Once the roster lands, this branch should go green with no change.

Blocked, and not by the change in this pull request. Leaving it open with the evidence rather than merging red or closing it. ## What CI says `ci / test` fails in `role-drift-check`: ``` a role's selection changed: agent/rendered/roles/devrel.bundle.txt, agent/rendered/roles/eval.bundle.txt, agent/rendered/roles/frontend.bundle.txt, agent/rendered/roles/gamedev.bundle.txt, agent/rendered/roles/platform.bundle.txt, agent/rendered/roles/sysadmin.bundle.txt, agent/rendered/roles/tpm.bundle.txt ``` Every name in that list is a role this repo has never recorded. ## Why it is not this change The recorded roster on `main` is the old naming: ``` ai creator design director engineer exec librarian ops qa ``` A bake against the current catalog produces the Core Roster v2 seats instead: `devrel`, `eval`, `frontend`, `gamedev`, `platform`, `sysadmin`, `tpm`. That is an upstream rename, and nothing in this branch touches roles. The branch changes `AGENTS.md` and nothing else. `scripts/task.sh` clones the catalog at `${AOS_CATALOG_REF:-main}`, unpinned, so the drift check compares this repo's committed record against whatever agentic-os `main` holds at the moment CI runs. agentic-os `main` took the seat retitles, so the record went stale without a commit here. `main` is green only because it has not re-run since. Any pull request opened in this repo right now fails the same way, this one included. ## What it needs A rebake and re-record of the new roster, which is the seat-rename migration rather than a drive-by on a docs change. I did not do it: the bake is only as correct as the catalog it reads, and recording a roster from an arbitrary local checkout would bake unrelated upstream drift into this repo's committed record. Filed the unpinned-ref half separately. Once the roster lands, this branch should go green with no change.

coilyco-ops referenced this pull request

2026-08-26 02:55:38 +00:00

role-drift-check clones the catalog at an unpinned main, so an upstream merge turns every pull request red with no commit here #1157

coilyco-ops referenced this pull request from coilyco-flight-deck/infrastructure

2026-08-26 03:01:42 +00:00

feat(scripts): add the git-workflow block bulk migration #943

coilyco-ops referenced this pull request from coilyco-flight-deck/infrastructure

2026-08-26 03:07:58 +00:00

AGENTS.md tells agents the merge is director-gated, which inverts the pull-request-and-merge lane and strands finished work #937

coilyco-ops added 1 commit

2026-08-27 05:08:10 +00:00

Merge remote-tracking branch 'origin/main' into chore/git-workflow-block

ci / image-build (pull_request) Successful in 59s

Details

ci / test (pull_request) Successful in 1m25s

Details

ci / publish-echo-image (pull_request) Has been skipped

Details

ci / publish-observed (pull_request) Has been skipped

Details

bc059a5038

coilyco-ops commented

2026-08-27 05:14:36 +00:00

Author

Owner

Superseded by #1163, which merged as 1a63704.

Both branches render the same managed Git workflow block into AGENTS.md. They differ only in which lane they render it for. This one renders pull-request-and-merge and keeps the frontmatter as it was. #1163 was filed a day later, after Kai settled the question, and declares pull-request because this repo gates the merge.

The two cannot both land: the second would overwrite the first with the wrong lane. #1163 carries the decided answer, so this one has nothing left to add and now conflicts with main.

Closing as superseded rather than merging. No work is lost, and nothing here needs redoing.

Superseded by #1163, which merged as 1a63704. Both branches render the same managed `Git workflow` block into `AGENTS.md`. They differ only in which lane they render it for. This one renders `pull-request-and-merge` and keeps the frontmatter as it was. #1163 was filed a day later, after Kai settled the question, and declares `pull-request` because this repo gates the merge. The two cannot both land: the second would overwrite the first with the wrong lane. #1163 carries the decided answer, so this one has nothing left to add and now conflicts with `main`. Closing as superseded rather than merging. No work is lost, and nothing here needs redoing.

coilyco-ops closed this pull request

2026-08-27 05:14:44 +00:00

ci / image-build (pull_request) Successful in 59s

Details

ci / test (pull_request) Successful in 1m25s

Details

ci / publish-echo-image (pull_request) Has been skipped

Details

ci / publish-observed (pull_request) Has been skipped

Details

Pull request closed

Please reopen this pull request to perform a merge.

Rows
Columns

docs(agents): render the managed AGENTS.md git-workflow block #1156

What CI says

Why it is not this change

What it needs

Pull request closed