Contracts
boundary-decision.md
Boundary Decision: Contract-Conformance Overlap (FR-005)
Mission: ci-local-preflight-parity-01KWXWY0 (#2283 Phase 3) Status: Adjudicated Owning requirement: FR-005 / SC-005
Context
#2283's original premise named three overlapping "did a retired contract break something" mechanisms: a dynamic review-time gate, a static grep-absence sweep, and the pre-existing stale_assertions post-merge analyzer. Three independently-evolved mechanisms catching slices of the same failure mode is the over-scope trap this mission is explicitly constrained to avoid (C-002: ship no 4th detector, no new allowlist artifact). This mission does not build a mechanism — it adjudicates which existing/incoming mechanism owns which slice, and files the one durable gap underneath all three.
Decision
1. Factor (c-dynamic) — #2438, CONDITIONAL on merge
#2438 (feat(review): auto-scoped review-time regression gate at move-task --to for_review, M5's pre_review_gate.py) discharges factor (c-dynamic) — "a consuming test now fails" — by re-running the consuming shards at for_review time.
This is stated conditionally, not as settled fact: pre_review_gate.py is NOT yet on the target branch (feat/ci-delivery-topology) — #2438 is still an open, unmerged PR at the time of this decision. The correct status is:
> (c-dynamic) is delivered-pending-#2438-merge, NOT closed.
This mission does not rebuild pre_review_gate.py, does not double -implement its selection logic (C-003), and does not treat #2438's landing as guaranteed. Whoever verifies #2283's factor (c-dynamic) closeout after this mission MUST re-check that #2438 has actually merged to upstream before treating that clause as discharged.
2. Factor (c-static / c′) — owned by CT7 (#2077), sharpened payload
Factor (c-static/c′) — "a production caller still references a retired contract that NO test exercises" — is a genuine residual that a test run structurally cannot catch (a test run proves the presence of a new failure, never the absence of a retired symbol). It remains open and is owned by CT7 (#2077) ("CT7: Test-hygiene directive/styleguide + recurrence -prevention guard").
The handoff payload is sharpened here, not left as a vague pointer:
> *Mechanise the test_no_legacy_ grep-absence family into a > content-anchored, allowlist-free retired-contract sweep, triggered on > shared-contract retirement.**
Today that family (tests/architectural/test_no_legacy_terminology.py, tests/architectural/test_no_legacy_status_emit_callers.py, tests/audit/test_no_legacy_agent_profiles_path.py, tests/audit/test_no_legacy_path_literals.py) is four independently hand-written sweeps, one per retirement, with no shared machinery. CT7's job is to generalize that pattern into one reusable, content-anchored (never line-pinned), allowlist-free primitive that fires whenever a shared contract is retired — not to invent a fifth bespoke sweep here (C-002).
3. The durable missing boundary — filed and embedded
Underneath both of the above sits the actual durable gap: a shared contract and its retirement are not a modeled, owned artifact with a declared consumer set. #2438 (dynamic), stale_assertions (post-merge AST analysis), and the test_no_legacy_* family (static grep-absence) are three independently -scoped point solutions that each rediscover this same underlying absence — none of them is wrong, but none of them is the durable fix either.
A contract-ownership boundary issue has been filed to track this gap as its own tracked item, parented under #2283 and #2077:
> Filed issue: #2441 — Contract-ownership boundary: shared contracts + > their retirement are not a modeled, owned artifact
This mission ships no (c) mechanism code and no new allowlist artifact (C-002) — the mechanisation is CT7's (#2077), and the ownership-model gap is #2441's.
Summary table
| Factor | Mechanism | Status | Owner |
|---|---|---|---|
| (c-dynamic) | #2438 pre_review_gate.py | delivered-pending-#2438-merge, NOT closed | M5 / #2438 |
| (c-static / c′) | test_no_legacy_* → mechanised sweep | Open | CT7 (#2077) |
| durable boundary | contract ownership + consumer-set model | Filed | #2441 |
Non-goals (C-002)
This decision record does not introduce, and this mission does not ship:
CT7's, tracked at #2077.
- A new (c) sweep/grep mechanism.
- A new standalone static-analysis allowlist artifact.
- Any change to
pre_review_gate.pyor #2438. - Any change to the
test_no_legacy_*family itself — that mechanisation is