Why: Users must see truthful first replies when work queues or cannot start.
Evidence: RED and GREEN HTTP browser receipts plus Frontend 269 tests and production build.
Impact: Adds two task-owned Playwright scenarios; production UI is unchanged.
Refs: U3 r0.5 independent review finding
Why: The first reply and task reuse need end-to-end checks across long and small material inputs.
Evidence: Dev evidence receipts for ten Playwright scenarios, Frontend tests, and build.
Impact: Adds browser-only coverage; production UI and API behavior are unchanged.
Refs: U3 r0.5 implementation plan Task 4
Why: One system reminder can contain multiple failed child notices, so the browser test must wait for the whole batch rather than one message per child.
Evidence: The first post-merge HTTP/Playwright run failed only at this false one-to-one assumption; all six task-owned browser cases passed after counting notice entries in the persisted reminder.
Impact: Test-only correction; no production UI behavior change.
Refs: U3 r0.4 Task 6 review closure
Why: Keep the approved RDM export controls and the U3 material browser coverage in one component candidate.
Evidence: Changes in store.ts touch separate state fields and behavior; full Frontend tests/build and task-owned browser cases will be rerun after this merge.
Impact: Includes upstream RDM export UI without changing U3 source or notice contracts.
Refs: Dev U3 r0.4 Task 6 final review synchronization
Why: A failed task card can appear before its notification turn, and the second finding must be observed while work is still running.
Evidence: The strengthened browser checks first exposed missing fixture notice text and a zero-width second-stage window; after fixture correction all six task-owned HTTP/Playwright scenarios passed. Frontend 265 unit tests and build passed.
Impact: Test coverage only; no production UI change.
Refs: U3 r0.4 independent review Important 5
Why: The r0.4 acceptance must show running findings, zero and partial failure, and a source-bound choice in the real UI.
Evidence: Six task-owned HTTP/Playwright cases passed; npm test (265 tests) and npm run build passed.
Impact: Browser test coverage only; no production UI change.
Refs: U3 r0.4 Task 6
Why: The U3 browser gate must show a host-generated pending question for underdetermined source counts.
Evidence: The new assertion failed against the old mechanism fixture while existing source-navigation and focus-flow tests remained observable; Backend fixture now emits structured premises.
Impact: Browser test inputs and assertion only; no production UI change.
Refs: Backend 756679d; Dev task feature/fast-slow-u2-u3-plan
Why: The Core mirror must satisfy the Frontend unused-export guard without changing the candidate shape.
Evidence: Dev make contracts regenerated the file; Frontend knip 51 <= 52 and build passed.
Impact: Type-only export visibility change.
Refs: Core b474394; Dev task feature/fast-slow-u2-u3-plan
Why: Frontend generated types must follow the single Core checkpoint contract.
Evidence: Dev make contracts generated the mirror; npm test and npm run build passed.
Impact: Type-only mirror update; persisted task cards and reading behavior stay unchanged.
Refs: Core c722c16; Dev task feature/fast-slow-u2-u3-plan
Why: Preserve current main before publishing the U2/U3 candidate.
Evidence: Exact main object fetched; browser-only upstream changes.
Impact: Retain both sets of browser coverage.
Refs: Agent-System-Dev U2/U3 delivery
Why: Material records now display validated source quotations rather than unverified model paraphrases.
Evidence: Existing browser coverage retains refresh, changed-version, preview and focus checks; the updated assertions target actual source text.
Impact: Browser-only expectations change; product bundle remains the independently reviewed version.
Refs: Approved R05 and host material-record guard
Why: Late source requests must not replace a newer preview or survive conversation navigation.
Evidence: Frontend typecheck, 221 unit tests, knip ratchet and build passed; delayed-response regressions reproduced the old failure.
Impact: The existing store owns source requests and invalidates stale responses on preview and context changes.
Refs: Agent-System-Dev feature-fast-slow-u2-u3-plan; independent review finding 4
Why: Unrelated browser suites should skip this opt-in runtime acceptance instead of importing an absent manifest.
Evidence: Playwright discovers the two tests without an instance; explicit owned-instance replay passed previously.
Impact: Preserves strict manifest verification when invoked.
Refs: fast-slow-u2-u3 Task 7
Why: DOM fixtures alone cannot verify persisted checkpoints and source version handling.
Evidence: SELF source navigation and changed-version checks pass; reading replay is being repeated after the deferred runtime port fix.
Impact: Adds an explicit task-owned instance browser suite without paid model calls.
Refs: fast-slow-u2-u3 Task 7
Why: Source context must remain attached to the bytes being read.
Evidence: SELF frontend tests and production build pass.
Impact: Carries an optional expected hash and preserves one preview location.
Refs: fast-slow-u2-u3 Task 5 source-version acceptance
Why: A folded auxiliary panel must not hide material limits or withdrawn conclusions.
Evidence: SELF 213 frontend tests, type checks, knip ratchet, and bundle build pass.
Impact: Prioritizes existing task cards without opening files or changing focus.
Refs: fast-slow-u2-u3 Task 5
Why: Replayed events, snapshot races, and source navigation must preserve authoritative work and user focus.
Evidence: SELF make evidence records 213 frontend tests, type checks, knip ratchet, and production build.
Impact: Uses the existing store for task revisions and gap replay; opens sources in the existing preview with version context.
Refs: fast-slow-u2-u3 Task 5; generated public contract types account for the measured knip baseline increase
agent-exec-transport
weichen