Home / Projects / Orchestrata Ultimate / Decisions / ADR index / ADR-0130
ADR-0130: ADRs are a publish step — rulings live in the scratch notes while a thread is open and are minted together at its end
| field | value |
|---|---|
| Status | Accepted — Accepted (Mohamed, 2026-10-22: "adr becomes a publish step for you") |
| Date | 2026-10-23 |
| Register | factory (docs/adr, 4-digit ids) |
| Source | docs/adr/0130-adrs-are-a-publish-step-rulings-live-in-the-scratch-notes-while-a-thread-is-open-and-are-minted-together-at-its-end.md — recorded on branch fix/ship-sanctioned-gate-pytest at 8c7031c9a, not yet on master |
| Wiki register entry | ADR-0130 (wiki status: Accepted) |
The record
Status: Accepted (Mohamed, 2026-10-22: "adr becomes a publish step for you") · Date: 2026-10-23
Context
The decision skill, following ADR-0046/0047, says that a ruling is recorded the same turn as an ADR in its three homes, with the drift gate green. On 2026-10-22, during the components cut-over review, rulings arrived several per hour, and some were later refined. Minting each one at once meant ADRs written mid-research and superseded soon after, plus repo writes in the middle of a discussion.
Mohamed's sequence that day:
- "stop writing in files. keep a scratch file"
- "hold them for a publish step i am not done", about two ADR-0124 edits already on disk
- "not an adr", on minting the test-runner ruling
- "when we have finished understanding, researchimg and agreeing adr becomes a publish step for you"
Decision (Mohamed, verbatim: "when we have finished understanding, researchimg and agreeing adr becomes a publish step for you")
- While a thread is still being understood, researched and agreed, each ruling is recorded verbatim in the session scratch notes, not as an ADR. That record satisfies the decision hook's "record it".
- At the end of the thread, or when Mohamed says publish, the scratch rulings are reviewed against
the register.
- A ruling already covered becomes an addendum on the existing ADR, where it is not Accepted, or is noted as covered.
- An operational pick goes to the status document.
- Each remaining ruling is minted, in docs, wiki and graph, with the gate green and both repos committed.
- Questions still open at publish time are minted as Proposed, with options and one recommendation.
Alternatives rejected
- Mint every ruling the same turn. ADRs get minted mid-research and then superseded, and the repo is written to during the discussion. Mohamed rejected this: "not an adr", "stop writing in files".
- Never write the rulings anywhere until publish. Rulings get lost across context compaction. The scratch file is the durable interim record.
Consequences
- The decision skill's "record the same turn" is narrowed: the scratch note is the same-turn record, and the ADR is the publish step.
- Risk: the scratch notes live under the session's
/tmpscratchpad and can be lost with it. The publish step must happen before the session ends, and a session handoff names any unpublished rulings. - First publish: 2026-10-23 (ADR-0127..0133, plus the ADR-0124 addendum), on Mohamed's "publish them to the wik and review the decisions in the scratch pad, update them and publish them if they havent already been covered".
Provenance
The quotes above are from Mohamed, 2026-10-22 and 2026-10-23. Memory file
adr-is-the-publish-step.md. The session scratch notes notes.md hold the rulings published here.
Related ADRs
Rulings cited
None.
Specs cited
None (or no spec page exists for the ones named).