core/docs/decisions/ADR-0103-fluency-lane-attachment-for-adr-0102.md
Shay 1395ec1354
feat(packs): ADR-0103 — attach hebrew_fluency + koine_greek_fluency lanes to ADR-0102 (#106)
* feat(evals): add Hebrew fluency holdout cases

* feat(evals): add Koine Greek fluency holdout cases

* feat(packs): attach fluency lanes to he_core_cognition_v1

* feat(packs): attach fluency lanes to he_logos_micro_v1

* feat(packs): attach fluency lanes to grc_logos_cognition_v1

* feat(packs): ADR-0103 fluency lane attachment

* test(packs): expect ADR-0103 fluency lanes on Hebrew Greek contracts

* docs(evals): add Hebrew fluency holdout split note

* docs(evals): add Koine Greek fluency holdout split note

* docs(evals): note Hebrew holdout attachment

* docs(evals): note Koine Greek holdout attachment

* docs: add ADR 0103 placeholder

* docs(adr): expand ADR-0103 fluency lane attachment

* docs: index ADR-0103 and refresh frontier
2026-05-22 09:43:46 -07:00

2.1 KiB

ADR-0103 — Fluency Lane Attachment for ADR-0102

Status: Accepted Date: 2026-05-22 Author: CORE agents + reviewers Depends on: ADR-0091, ADR-0102


Context

ADR-0102 ratified hebrew_greek_textual_reasoning with only universal reasoning lanes attached:

  • inference_closure
  • fabrication_control

At the time, the language-specific Hebrew and Koine Greek fluency lanes did not yet expose sealed holdout splits, so attaching them to ratified contracts would have violated ADR-0091 expectations around dev/public/holdout coverage.

Both lanes now ship plaintext holdout sets through:

holdouts/v1/cases.jsonl

and can therefore be attached to the ADR-0102 manifests without weakening the contract discipline.


Decision

Attach the following eval lanes to all four ADR-0102 manifests:

  • hebrew_fluency
  • koine_greek_fluency

Each lane declares:

{
  "version": "v1",
  "splits": ["dev", "public", "holdout"]
}

The following packs must remain uniform:

  • he_core_cognition_v1
  • he_logos_micro_v1
  • grc_logos_cognition_v1
  • grc_logos_micro_v1

All four manifests advance provenance to:

adr-0103:reviewed:2026-05-22

Acceptance evidence

Accepted after the following landed together:

  • Hebrew holdout cases
  • Koine Greek holdout cases
  • updated fluency contracts
  • updated gap notes
  • updated sibling-ratification invariants
  • uniform lane attachment across all four manifests

Invariants

hebrew_greek_fluency_lanes_uniform

All four ADR-0102 packs must declare identical eval-lane coverage.

hebrew_greek_fluency_holdouts_present

Both:

  • hebrew_fluency
  • koine_greek_fluency

must ship non-empty holdouts/v1/cases.jsonl before appearing in ratified domain contracts.


Consequences

  • ADR-0102 now carries language-specific fluency evidence in addition to universal reasoning evidence.
  • Hebrew/Greek textual reasoning contracts now satisfy full dev/public/holdout discipline for attached fluency lanes.
  • Future language-track ADRs can extend morphology and construction coverage without changing Domain Pack Contract v1.