- ADR-0108 Status: Proposed -> Accepted - README index row updated to Accepted - 'Current frontier' rewritten with the ranked Proposed-ADR list mandated by ADR-0108 \xc2\xa7Decision; removes the now-false 'No ADR currently sits in a "Proposed but unimplemented" state' sentence - Open candidate directions (no-ADR-yet) section retained for the multi-reviewer governance frontier item from ADR-0105
5.4 KiB
ADR-0108 — Proposed-ADR Sequencing Post-ADR-0105
Status: Accepted Date: 2026-05-22 Author: CORE agents + reviewers Relates to: ADR-0080, ADR-0084, ADR-0087, ADR-0106
Context
After ADR-0105, the ADR index carries four Proposed-but-not-Accepted ADRs:
| ADR | Proposed | Origin |
|---|---|---|
| ADR-0080 | 2026-05-20 | Contemplation loop (self-interrogation, read-only Phase 1) |
| ADR-0084 | 2026-05-20 | Definitional layer for lexicon packs (optional per-entry block) |
| ADR-0087 | 2026-05-20 | Rhetorical style as a third substantive selection axis |
| ADR-0106 | 2026-05-22 | Expert-demo promotion contract |
Three of these (0080 / 0084 / 0087) were proposed in a single 2026-05-20 session and have not advanced. The frontier work since that day went into the ADR-0091→0105 evidence-governed domain chain (a different axis), so the stall is orthogonal scope, not contradiction or rejection.
docs/decisions/README.md currently states "No ADR currently sits in a
'Proposed but unimplemented' state." That sentence is no longer true and
the README will need updating once this sequencing decision is accepted.
Without an explicit sequencing decision, three things happen:
- The Proposed-ADR list grows without external legibility.
- Reviewers cannot tell intent to land from intent to defer from intent to withdraw.
- Future "what's next?" questions re-derive sequencing on every read, which is wasted work and drifts over time.
ADR-0108 makes the sequencing decision explicit, durable, and revisable.
Decision
1. Priority order
The four currently-Proposed ADRs are sequenced as follows:
-
ADR-0106 — Expert-Demo Promotion Contract. Highest priority. Domain-legibility: the four
reasoning-capableratifications overstate what the system has demonstrated untilexpert_demohas a real contract. Implementation PR follows acceptance. -
ADR-0107 —
mathematics_logicexpert-demo promotion (reserved). First worked promotion against the ADR-0106 contract. Smallest expert-demo proof surface across the four ratified domains. -
ADR-0080 — Contemplation Loop. Next-most-load-bearing. Converts gap-finding from human-driven to system-emitted-and-reviewed. Phase 1 is intentionally read-only and
SPECULATIVE-only. Unlocks curriculum growth speed. -
ADR-0084 — Definitional Layer. Deferred pending ADR-0107. The definitional block is content-shaped; its value surfaces during a worked expert promotion when definitional depth becomes a bottleneck. Holding it Proposed avoids premature schema commitment.
-
ADR-0087 — Rhetorical Style Axis. Lowest current priority. Register + anchor-lens already demonstrate the orthogonality pattern ADR-0087 generalizes. No active pull from a downstream consumer. Stays Proposed until a concrete consumer ADR motivates it.
2. No withdrawals
This ADR withdraws none of 0080 / 0084 / 0087. Each remains Proposed and intact. Sequencing ≠ rejection.
3. Sequencing is revisable
Acceptance of ADR-0108 does not freeze the order. The order is the current best guess; new evidence (a worked promotion that hits a definitional bottleneck, a downstream rhetorical-style consumer landing, an unanticipated capability gap) updates the order by a follow-up ADR.
4. README discipline
The README "Current frontier" section MUST list every Proposed ADR with its sequencing rank and one-line rationale, so outside readers can see intent without reading every ADR body. The sentence "No ADR currently sits in a 'Proposed but unimplemented' state" is removed when this ADR lands and is replaced with the sequenced Proposed-ADR list.
Invariants
proposed_adr_index_complete
Every ADR with Status: Proposed in docs/decisions/ must appear in the
README "Current frontier" sequencing list. A Proposed ADR absent from the
sequencing list is a documentation drift bug.
no_silent_withdrawal
A Proposed ADR may move to Withdrawn only by a successor ADR that
explicitly cites it. No silent deletion.
sequencing_is_revisable
Acceptance of ADR-0108 does not lock the order. A subsequent ADR may re-rank Proposed ADRs by citing this one and stating the trigger.
Acceptance evidence
Accepted when the following land together:
- README "Current frontier" updated to:
- remove the "No ADR currently sits in a 'Proposed but unimplemented' state" claim
- add the ranked Proposed-ADR list with one-line rationales matching §Decision
- this ADR's status flipped to Accepted
- no body content of ADR-0080 / 0084 / 0087 / 0106 is modified by this PR (sequencing is a meta-decision, not a content edit)
Consequences
- "What's next?" becomes a one-screen read on
docs/decisions/README.mdrather than a four-file scan plus reasoning. - Reviewers can act on Proposed ADRs in declared order without cross-checking session memory.
- Proposed ADRs that genuinely no longer fit get a documented exit via a successor ADR rather than disappearing from the index.
- Future "candidate frontier" bullets in the README become the on-ramp to a new Proposed ADR, with this sequencing list as their next stop.
Out of scope
- This ADR does not implement any of the four Proposed ADRs.
- This ADR does not modify ADR-0080 / 0084 / 0087 / 0106 bodies.
- Multi-reviewer governance (the open ADR-0105 candidate frontier item) is orthogonal and remains future work.