feat(agents): declare the git workflow in AGENTS.md frontmatter #1663

Merged

coilysiren merged 1 commit from aos/claude/mg96-fm into main

2026-08-14 05:17:17 +00:00

coilyco-ops commented

2026-08-14 05:13:09 +00:00

Member

Part of the coilysiren/inbox#365 teardown, which deletes .ward/ward.yaml from repos.

Moves agent.workflow into AGENTS.md frontmatter:

---
ward:
  workflow: pull-request-and-merge
---

The lane is doctrine about how this repo is worked, so it belongs beside the doctrine rather than in ward's config. Namespaced under ward: so it does not collide with harness-level frontmatter keys.

This one is not a no-op. Read before merging.

This is a PR-lane repo, one of five declaring pull-request-and-merge. Ward does not read AGENTS.md frontmatter yet, and cmd/ward/agent_workflow.go:32 sets defaultWorkflow = workflowDirectToMain (merge-remote-main).

So between this merging and coilyco-flight-deck/ward#1661 landing, a ward dispatch against this repo would resolve to merge-remote-main and push straight to main with no pull request.

Ward has been inert for weeks, so the hazard is dormant rather than live (Kai's call). ward#1661 is a blocker on ward resuming dispatch. The declaration here is doctrine in the meantime, and .ward/ward.yaml no longer carries the lane.

The 27 non-PR-lane repos in this sweep declared merge-remote-main, which is already the compiled default, so those are true no-ops. Only these five change resolved behavior while ward#1661 is open.

Mechanically checked

The edit refuses to run unless the workflow key is found and removed, the resulting YAML re-parses, commands and catalog are byte-identical afterward, and the new AGENTS.md stays under the 290-line / 26400-char documentation-layout caps counting the frontmatter itself.

Part of the coilysiren/inbox#365 teardown, which deletes `.ward/ward.yaml` from repos. Moves `agent.workflow` into AGENTS.md frontmatter: ```yaml --- ward: workflow: pull-request-and-merge --- ``` The lane is doctrine about how this repo is worked, so it belongs beside the doctrine rather than in ward's config. Namespaced under `ward:` so it does not collide with harness-level frontmatter keys. ## This one is not a no-op. Read before merging. This is a **PR-lane repo**, one of five declaring `pull-request-and-merge`. Ward does not read AGENTS.md frontmatter yet, and `cmd/ward/agent_workflow.go:32` sets `defaultWorkflow = workflowDirectToMain` (`merge-remote-main`). So between this merging and coilyco-flight-deck/ward#1661 landing, a ward dispatch against this repo would resolve to **merge-remote-main** and push straight to `main` with no pull request. Ward has been inert for weeks, so the hazard is dormant rather than live (Kai's call). **ward#1661 is a blocker on ward resuming dispatch.** The declaration here is doctrine in the meantime, and `.ward/ward.yaml` no longer carries the lane. The 27 non-PR-lane repos in this sweep declared `merge-remote-main`, which is already the compiled default, so those are true no-ops. Only these five change resolved behavior while ward#1661 is open. ## Mechanically checked The edit refuses to run unless the workflow key is found and removed, the resulting YAML re-parses, `commands` and `catalog` are byte-identical afterward, and the new AGENTS.md stays under the 290-line / 26400-char documentation-layout caps counting the frontmatter itself.

coilyco-ops added 1 commit

2026-08-14 05:13:10 +00:00

feat(agents): declare the git workflow in AGENTS.md frontmatter

test / test (pull_request) Successful in 1m10s

Details

0851bd559d

Moves agent.workflow out of .ward/ward.yaml, part of deleting that file
(coilysiren/inbox#365). The lane is doctrine about how this repo is worked,
so it belongs beside the doctrine rather than in ward's config.

Ward does not read the new location yet (ward#1661). It resolves the lane
from ward.yaml and falls back to merge-remote-main when absent.

Refs coilysiren/inbox#365, ward#1661

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Kai Siren <coilysiren@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>

coilyco-ops requested review from coilysiren

2026-08-14 05:13:10 +00:00