Uh oh!
There was an error while loading. Please reload this page.
chore: retire the ycst import blocks and record the migration - #68
Conversation
There was a problem hiding this comment.
Retires the spent imports.tf now that the ycst-org-uk migration is applied, and records the migration + two criteria corrections in decisions/ and quality/criteria.md. Purely documentation and cleanup — no module or data changes, and the PR body's claim that task plan reports "No changes." is consistent with the diff. Cross-checked the new decision record and criteria amendments against decisions/2026-08-13-team-member-roles.md, the referenced design/plan docs, and the existing data/ycst-org-uk/*.yaml files; all references resolve and the technical claims (moved cannot carry a rename because github_repository's ID is the repo name, removed rejects instance keys) are accurate and consistent with the deleted imports.tf's own commentary. No Critical or Important findings.
Uh oh!
There was an error while loading. Please reload this page.
Closes out the
ycst-org-ukmigration. No plan-affecting change —task planreports
No changes.withimports.tfdeleted, which is the point of the PR.Deletes the spent
imports.tf, records the migration indecisions/, andamends two
quality/criteria.mdcriteria that this work proved incomplete.The
moved/removedcriterion said instance address changes producedestroy+create because Stategraph ignores those blocks. That described the
retired
stategraph tfwrapper. Under the native CLImovedis honoured,including across provider aliases and renamed
for_eachkeys — but it cannotcarry a rename, because
github_repository's ID is the repo name. Terraformrewrites the address and the provider binding, never the ID, so refresh looks
for the old name under the new owner and 404s.
The create-vs-exists criterion said a plan proposing to create resources
that already exist means a misconfigured backend. That is one of two causes; the
other is a resource ID that no longer resolves, and the two are indistinguishable
in the plan. The backend was fine here.
Last triggeredalso records the second firing of the credential criterion — a${VAR:+set}guard undone by a${VAR:-...}fallback on the same line printed aPAT in full. Token rotated. A third firing promotes it to an automated check, so
there is a suggested hook pattern in the file.
Also notes on
decisions/2026-08-04-ycst-admin-docs-private-cpanel.mdthat therepo now lives at
ycst-org-uk/board-docs, and marks the design doc implementedwith a pointer to what its finding 1 got wrong.
🤖 Generated with Claude Code
https://claude.ai/code/session_01C15usdmazQC71xZC1YFGHg