[bokelley]

Adopts the AdCP 3.3.0-beta.4 reporting-subscription changes (adcontextprotocol/adcp#8718, #8717, #8724) on the SDK 10 / next line.

Three commits:

1. feat(reporting): the coverage, grant and controller rules.
2. fix(reporting): the TS-parity harness fixes.
3. feat(protocol)!: the signed bundle pin, from schemas/releases/3.3.0-beta.4.json. The source commit is 8ee1bfa2e9e7, taken from the Fulcio certificate, which is also the commit the vendored coverage vectors come from. Bundle sha256 3525423d25bc…, cosign-verified. The schema cache, packaged compliance tree, request-signing vectors, generated types and public API snapshot are all regenerated.

Changes

1. Per-package added_at (rule change). SubscriptionBuyPackage.added_at is read from get_media_buys packages[].added_at, and the managed path keeps it, also under buy_offerings. SubscriptionBuy.package_confirmed_at() returns max(confirmed_at, added_at). A period is owed when it ends after that instant for at least one package resolving the offering. Each package resolves at the later of the version's accepted_at and its own instant, and adding a package is never a cutover. explain has the new before_package_added reason, with the same facts (packages_added_at), sentence and buy facts as the TS SDK. The test kit gains kit.buys.add_package, which serves added_at.
2. Reduced grant without get_reporting_status. The engine already treated this as a loss at the reduction (accounts_unauthorized, withheld periods, re-grant when the task is added back). It now keeps the reduced grant's tasks, so get_reporting_status refuses the still-visible account with SCOPE_INSUFFICIENT. A revoked account stays ACCOUNT_NOT_FOUND.
3. SUBSCRIPTIONS_REQUIRE_ACCOUNT_ID_KEY. check_subscription_offerings() runs the spec's offering-checks.json and passes all 8 cases. The managed service now refuses a managed-delivery offering whose primary_keys omit account_id; beta.3 only logged it. ReliableReportingConfigurationError.context carries offering_id and missing.
4. Schema changes (regenerated):
   • get_media_buys gains Package.added_at, and the controller response gains SetAccountTimezoneResult.
   • Mixed date and time durations under a local alignment are now wire-invalid, so the local-duration-cannot-mix-date-and-time vector checks that.
   • DestinationOverride is still one model with all four fields (account_ids, operator_domains, destination_id, path_template).
   • Numbered names renumber (Status54 → Status53, ...).
   • create_media_buy drops Opportunity / Status (adcp#8709).
   • The schema-parity test synthesizer now merges a composition branch's own property schemas with the root's.
5. tzdata. The vectors' new zones (Pacific/Auckland, Pacific/Chatham, Asia/Kolkata, America/Mexico_City) pass under the existing pin, tzdata==2026.6 (2026f). The parity test ignores the host's zoneinfo. No runtime pin: zoneinfo prefers the system database over the tzdata package, so pinning would not make results deterministic. The vectors' dates also agree across tzdata 2026b through 2026f. This is documented in docs/reporting-subscriptions-coverage.md.
6. Controller.
   • reset_state is a scenario and a TestControllerStore hook. register_test_controller and the A2A server refuse a store that implements reporting_subscription_probe without it.
   • Probe results are nested under simulated, as SimulationSuccess requires.
   • The reference seller serves reset_state. It unpins the clock, restores the caller's default grant and the original timezone, and discards what the probe recorded via ReportingSubscriptionEngine.discard_sandbox_changes.
   • The reference seller's set_account_timezone returns SetAccountTimezoneResult.
   • The Python SDK has no storyboard runner implementing owed_periods_match, so the new window rule has nothing to apply to.

Parity fixes, matching adcontextprotocol/adcp-client#3620 (separate commit):

• A package_ids exception no longer changes resolution.
• excluded_by_buyer applies in-place membership only to the versions it overlaps.
• The 3 differential regressions are vendored and pass.
• At an account-timezone cutover, a billing feed owes a stub, following the cutover text. The 8ee1bfa2 reference model emits none there; the spec fixes that for beta.5.

Breaking

• The SDK speaks 3.3-beta.4 and no longer accepts a 3.3-beta.3 pin.
• A managed-delivery offering without account_id in primary_keys is refused.
• SCOPE_INSUFFICIENT replaces ACCOUNT_NOT_FOUND after a reduction that drops get_reporting_status.
• A TestControllerStore with the probe must implement reset_state.
• Probe results are nested under simulated.

These are listed in the MIGRATION_v9_to_v10.md and docs/adcp-3.3-sdk-inventory.md section 14.

Vectors

All 138 canonical coverage vectors pass, including the 18 not in the previously vendored file. Also passing: the 8 offering checks and the 3 differential regressions.

Storage contract

Unchanged. added_at is read from the host's buys on each computation and never persisted. The reduced grant's tasks use the existing optional allowed_tasks key of the engine state document (STATE_DOCUMENT_VERSION 1). No ledger column changes, so there is no REPORTING_STORAGE_CONTRACT bump and no migration.

🤖 Generated with Claude Code

[aao-secretariat]

Ladon verdict: Approve

Approve — AdCP 3.3.0-beta.4 adoption, clean review, no blocking findings.

The reviewer completed a thorough pass with zero critical/high/medium findings:

• Semver signal is correct: breaking adoption ships under feat(protocol)! with a BREAKING CHANGE footer and migration notes.
• Largest non-generated file (coverage.py, 183 changed lines) was studied: the new per-package owing filter (owing = regime.owing_packages(...), L1307-1314) is safe — the join-instant period stays owed in full and package addition is never a cutover, matching the TS SDK.
• engine.py reduced-grant handling correctly distinguishes SCOPE_INSUFFICIENT vs ACCOUNT_NOT_FOUND; offering_checks.py and controller gating are sound.
• A new beta4 test file (371 lines) covers the added branches.
• No _generated-layer or domain-import violations, no forward-compat/alias/ergonomic edits, no credentials in metadata, no disabled tests.
• The storage-migration-test deletion seen in base..head is a tooling artifact (provided base is ahead of the true merge-base); the PR does not touch that file.

high_risk is true, but all matching paths are schemas/** additions (new generated schema cache files) — new files are normal scaffolding, not an escalation trigger (no (modified)/(deleted) entries). gated_paths is false; no author team gate. Only a single non-blocking Low (orphaned SubscriptionException.removes_package), omitted from the gate.

Decision table: no critical/high (row 1 n/a), gated_paths false (row 2 n/a), no (deleted) high-risk reasons (row 3 n/a), no medium findings (rows 4/5/8 n/a), no prior escalate (row 6 n/a), no team gate (row 7 n/a) → row 9 approve.

[aao-secretariat]

Ladon verdict: Approve

Approve — test-only delta with no blocking or medium findings.

This PR's delta is a single test-only commit: it deletes the discriminator_emitted entry from the ACCEPTED set in tests/conformance/test_wire_round_trip.py, leaving it empty. Per the file's own stated contract, the set only shrinks, and a class that no longer occurs under the beta.4 schemas must be removed; two tests gate the claim in both directions, so CI validates the change. The remaining high-risk flag entries are all schemas/** cache files marked (added) — new-file scaffolding, not escalation-worthy on the flag alone.

Decision table walk-through:

• Rows 1 (critical/high): no findings — skip.
• Row 2 (gated_paths): gated_paths is false — skip. (review_decision is also APPROVED.)
• Row 3 (deleted high-risk path): all high-risk reasons are (added), none (deleted) — skip.
• Row 4 (medium data-loss/schema/infra): no findings — skip.
• Row 5 (modified high-risk + medium): no (modified) entries and no medium findings — skip.
• Row 6 (sticky escalate): prior decision was approve, not escalate — skip.
• Row 7 (no-auto-approve team): no team match — skip.
• Row 8 (≥3 medium): zero medium findings — skip.
• Row 9: approve.

Prior pass was clean (approve, no findings); this subsequent run surfaces nothing new. Net clean.