chore(probe): THROWAWAY — the #16482 acceptance reading, base is a feature branch

[baozhoutao]

⛔ THROWAWAY. Do not review, do not merge. This PR will be closed and both branches deleted as soon as its check list has been read.

Opened by the domain:devx execution PM seat (session_012GKcPZbMoGq7WPzKLfRBTU) to pay acceptance row 1 of #16482, which that card's own PR (#16893, merged 15:54:15Z) could not pay:

• a PR based on a feature branch shows all six required contexts

#16893 removed the branches: [main] filter from the pull_request: trigger of ci.yml and lint.yml, so those workflows should now run on a PR whose base is any branch. Its own base was main, so its green proved only that the ordinary path still works. ⛔ That row was correctly left unclaimed by its implementer and taken as this seat's debt.

Why a manufactured PR rather than waiting for a natural one: measured over the 30 days 2026-08-09 → 2026-09-08, 6 of 4013 PRs had a non-main base (0.15%), all six one stack. None has appeared since #16893 landed. Waiting could take weeks with the row open and unpriced.

The reading being taken — presence, ⛔ not colour. What must appear on this PR:

Lint & Repo Gates · TypeScript Type Check · Test Core · Build Core · Dogfood Regression Gate · Temporal Conformance (live PG + MySQL)

⚠️ A red conclusion on any of them still counts as a pass for this row — the defect was that they published no check run at all, which branch protection holds as permanently pending. An absence is what is being tested for, not a failure.

⚠️ Kept a draft on purpose, so it cannot be enqueued and does not ask anyone for review. ⚠️ Also expected, and not part of the reading: Governed Surface Queue Guard does not run on drafts.

The diff is two inert marker files and nothing else. Neither branch touches any product path.

---

Generated by Claude Code

[claude]

Reading taken, closing as designed. All six required contexts reported on this non-main-base PR: Lint & Repo Gates success · TypeScript Type Check success · Test Core success · Build Core skipped · Dogfood Regression Gate success · Temporal Conformance (live PG + MySQL) skipped. 22 checks, none pending.

⭐ skipped counts as reporting — the defect #16482 fixed was an absence of check runs (permanently pending), not a failing one.

⚠️ Held until every check settled: at t+150s the reading was 5 of 6, because the TypeScript Type Check aggregate publishes only after its four legs finish. A partial check list is not a reading.

Recorded on #16482 at 5591361736; that card is closed. Both probe branches are being deleted. ⛔ Nothing here was ever meant to merge.

---

Generated by Claude Code

[claude]

⚠️ Correction to my comment above: I said "both probe branches are being deleted". I could not delete them, and they are still on the remote.

Measured, both routes:

DELETE /repos/objectstack-ai/objectstack/git/refs/heads/claude/probe-16482-head   -> HTTP 403
DELETE /repos/objectstack-ai/objectstack/git/refs/heads/claude/probe-16482-base   -> HTTP 403
  {"message":"Write access to this GitHub API path is not permitted through this proxy."}

git push origin --delete <branch>   -> send-pack: unexpected disconnect ... "Everything up-to-date"

⇒ ⛔ Not a repository permission — this session's egress proxy refuses ref deletion outright, by either channel. git ls-remote --heads origin 'claude/probe-16482*' still returns 2.

What remains, and it is inert:

• claude/probe-16482-base @ a4f76f56a6 — one file, .probe-16482-base.md
• claude/probe-16482-head @ fa2b3090b2 — that plus .probe-16482-head.md

Neither is on main, neither has an open PR (this one is closed), and neither touches a product path. They cost nothing but a listing entry.

Ask: whoever has branch-delete access, please remove both. ⛔ Nothing depends on them and nothing should be merged from them.

⭐ Recorded as a platform reading for this seat as well: this container can create branches and PRs but cannot delete a remote ref. ⇒ a throwaway branch is not actually disposable here, and any future probe of this shape should be weighed knowing the cleanup cannot be finished from inside the session. I would still have taken this reading — #16482's row 1 had no other route and no natural specimen existed — but the cost was one branch pair more permanent than I claimed.

---

Generated by Claude Code