browser-based guardian cockpit as the only supported browser shell
WebSocket conversation path
native macOS observer daemon for screen and OCR ingest
daemon screen analysis can now choose local Apple Vision, local codex exec, or explicit OpenRouter cloud OCR; local Codex writes only a temporary PNG for the CLI invocation and deletes it after analysis or failure unless local capture preservation is explicitly enabled
screen captures can be preserved as durable local artifacts for future re-analysis, pairing each allowed image with redacted provider output and normalized JSON behind localhost-only observer artifact endpoints
observer refresh pipeline across time, calendar, git, goals, and screen context
proactive delivery gating and queued-bundle delivery inside the current product
first coherent desktop presence surface built on daemon status, capture-mode visibility, pending native-notification state, a safe test-notification path, desktop-notification fallback when browser delivery is unavailable, browser-side controls for pending native notifications, a first learning-driven native-channel preference layer, one continuity snapshot that exposes daemon state, deferred bundle items, pending native notifications, and recent interventions together, plus action-card continuation payloads and an actionable cockpit desktop-shell card with dismiss/follow-up/continue controls
cross-surface continuity now also exposes explicit open-thread and continue flows across native notifications, queued interventions, and recent interventions
route-health snapshots now expose whether browser websocket and native delivery are actually reachable at runtime, including ready, fallback, and unavailable states with operator-readable repair hints instead of only static route bindings
queued-bundle native delivery now preserves same-thread resume when every deferred item belongs to the same session, and the shared continuity snapshot now carries one continuation contract across notifications, queued insights, and recent interventions
the shared continuity snapshot now also synthesizes continuity health, grouped follow-through threads, and explicit recovery actions, and the cockpit desktop shell now uses that richer contract instead of reconstructing cross-surface recovery from raw notifications, queued items, and route rows
the shared continuity snapshot now also carries imported capability-family attention and typed source-adapter degradation, so browser/native route health is no longer the only reach surface visible in observer continuity or cockpit recovery flows
broader reach continuity now also propagates into the operator timeline and Activity Ledger, so typed source-adapter and imported-reach recovery is not stranded in the observer endpoint or cockpit-only surfaces
observer continuity now also carries explicit presence-surface inventory across channel adapters, messaging connectors, node adapters, and observer definitions, with ready versus attention summaries, follow-up prompts, repair hints, and the same presence recovery surfaced into cockpit triage, the desktop shell, the threaded operator timeline, and the Activity Ledger
observer continuity now also carries inventory-backed browser-provider and node-adapter reach surfaces with selected-versus-fallback state, network or daemon prerequisites, and the same repair or follow-up recovery contract propagated into cockpit triage, the desktop shell, the threaded operator timeline, and the Activity Ledger
cross-surface continuity now also lands in an explicit operator continuity graph that ties notification and deferred-guardian follow-through back to session, workflow, approval, artifact, and intervention state instead of leaving the thread graph implicit across APIs
backend reach/integration proof now runs through isolated per-file backend shard execution instead of one shared shard-wide pytest process, reducing async teardown contamination in CI
browser and desktop execution proof now also has a dedicated benchmark lane: replayable browser-task receipts, desktop notification-action replay, cross-surface continuity, and operator-visible computer-use failure taxonomy now live in the named computer_use_browser_desktop suite plus the operator computer-use benchmark surface instead of staying spread across isolated continuity or daemon seams
M4 channel proof now also has a named deterministic benchmark lane, channels_presence_device_pairing, covering channel identity boundaries, device pairing and revocation fail-closed visibility, external-channel continuity, mutation boundaries, and abuse/failure review receipts; this is benchmark proof of boundary posture, not a claim that broad live mobile or messaging transports are production-ready
Batch BY reach/browser/voice hardening now also has named deterministic lanes: production_reach_channel_hardening, browser_computer_use_reliability_v2, and guardian_safe_voice_media_runtime, plus /api/operator/production-reach-browser-voice, covering paired external messaging identity, revocation, approval handoff, privacy redaction, degraded recovery, browser provider truth, session partitioning, crash recovery, page-drift replay blocking, and guarded voice/media correction/deletion/revocation paths without claiming broad live reach, voice parity, or safe browser automation
Batch CE live broad reach and production voice/media proof now adds live_broad_reach_channel_attestation, production_voice_media_provider_runtime, cross_surface_continuity_recovery, plus /api/operator/live-reach-media-proof, covering live/recorded-live mobile-push and messaging provider receipts, evidence mode, consent, pairing, revocation, rate limits, abuse handling, approval handoff, STT/TTS/media-analysis capture boundaries, correction/deletion, provider-failure fallback, and cross-surface thread/memory/approval continuity without claiming broad reach, complete channel coverage, OpenClaw-class reach, voice parity, multimodal parity, production STT/TTS solved, production mobile execution solved, always-available operation, production readiness, or full parity
Batch CL broad reach, production voice/media, and mobile execution proof now adds broad_channel_sla_operations, production_voice_media_quality_gates, mobile_execution_continuity, plus /api/operator/production-reach-voice-mobile, covering provider windows, rate-limit and abuse handling, degraded recovery, coverage-gap claim boundaries, STT/TTS/media quality and latency gates, correction/deletion privacy boundaries, provider-regression fallback, notification approval handoff, mobile action continuity, thread/memory recovery, offline recovery, and revocation fail-closed behavior without claiming OpenClaw-class reach, voice parity, multimodal parity, always-available operation, production readiness, or full parity
Batch CU broad-reach field operations and voice/media quality operations now add broad_reach_field_operations, voice_media_quality_operations, always_available_reach_slo, plus /api/operator/broad-reach-field-ops, covering provider/channel field matrices, consent and revocation fail-closed checks, abuse/rate-limit drills, degraded recovery, continuity IDs, metadata-only redacted receipt boundaries, voice/media quality and latency gates, correction/deletion/privacy controls, bounded SLO windows, provider-failure and offline recovery, coverage gaps, and blocked claims without claiming OpenClaw-class reach, complete channel coverage, voice parity, multimodal parity, always-available operation, production readiness, or full parity
Batch DC selected-channel reach/media campaign proof now adds always_available_reach_operations_v1, voice_media_parity_runtime_v1, mobile_cross_surface_continuity_v1, reach_degraded_recovery_field_campaign, plus /api/operator/always-available-reach-media, covering selected mobile/messaging/native/browser/web campaign windows, pairing/revocation, rate-limit and abuse recovery, 14-day equivalent degraded-recovery metrics, voice/media quality and latency receipts, correction/deletion/privacy controls, provider-regression fallback, cross-surface continuity, false/missed delivery metrics, operator repair, redacted receipts, and blocked claims without claiming OpenClaw-class reach, complete channel coverage, always-available operation, voice/media parity, production readiness, or full parity
Batch CH browser-provider usability proof now adds managed_browser_provider_attestation, live_multi_operator_usability_study, browser_computer_use_recovery_drill, plus /api/operator/browser-provider-usability-proof, covering local/managed/remote browser provider identity, evidence mode, session partitioning, credential and download/upload boundaries, provider degradation, recorded-live multi-operator usability metrics, and fail-closed browser recovery drills without claiming safe autonomous browser/computer-use, full browser parity, best cockpit, solved operator control, production readiness, or full parity
the one excellent reach-channel canary now selects native notifications as the only canary channel and exposes one_excellent_reach_channel_canary, /api/operator/one-reach-channel-canary, and the cockpit benchmark-proof card for pairing, revocation, health, retry/fallback, same-thread continuity, memory/context continuity, approval handoff, audit trail, degraded-state UI, and explicit rejection of Slack/Discord/Telegram/channel sprawl until one channel meets the bar
Seraph can observe and update state outside the immediate chat loop
Seraph can proactively surface output in the current product
Seraph has at least one real non-browser presence path outside the browser tab
Seraph now has a coherent first desktop presence surface rather than only a browser app plus hidden daemon fallback
browser and native pending-notification state now have a first shared control surface
browser and native continuity now share one operator-readable snapshot for daemon, notification, queued-bundle, and recent intervention state
browser and native continuity now also share one explicit continuation contract plus runtime route-health state instead of reconstructing thread or fallback semantics per surface
the cockpit now exposes a first actionable desktop shell surface for pending alerts, queued items, and recent interventions
native continuation payloads can now resume work back into the cockpit instead of only showing a passive notification
M4 benchmark proof names identity, pairing, revocation, mutation, external-continuity, and abuse/failure review boundaries for channel/device reach
the native-notification canary now proves one selected reach channel with explicit no-sprawl boundaries, approval handoff, memory/context continuity, retry/fallback, and degraded-state receipts
guardian-safe multimodal and voice now has a dedicated proof gate: guardian_safe_multimodal_voice and /api/operator/guardian-safe-multimodal-voice require voice, TTS/STT, browser vision, image/media analysis, and media delivery families to expose owner, trust, permissions, capture/provider/privacy receipts, continuity, correction/deletion, revocation, and guardian-value reasons before being treated as reach
Batch BY adds guardian_safe_voice_media_runtime receipts for guarded voice/STT and browser-vision runtime paths with guardian-value reasons, transcript or media-summary privacy, operator correction, deletion, and revocation fail-closed checks
Batch CE adds live/recorded-live reach/media receipts for mobile push, external messaging, STT, TTS, media analysis, consent, pairing, revocation, provider failure, correction/deletion, and cross-surface thread/memory/approval continuity
Batch CU adds broad-reach field-operation, voice/media quality-operation, and bounded SLO receipts for provider/channel field matrices, consent/revocation fail-closed checks, abuse/rate-limit drills, degraded recovery, continuity IDs, metadata-only redacted receipt boundaries, voice/media correction/deletion/privacy controls, provider-failure/offline recovery, coverage gaps, and blocked claims
Batch DC adds selected-channel reach/media campaign, voice/media runtime, cross-surface continuity, and degraded-recovery field-campaign receipts for pairing/revocation, rate-limit and abuse recovery, 14-day equivalent windows, false/missed delivery metrics, operator repair, redacted receipt boundaries, and blocked claims
Batch CH adds browser-provider attestation and recovery receipts for local/managed/remote provider identity, evidence mode, session partitioning, credential/download/upload boundaries, provider degradation, and fail-closed recovery drills
broad always-available mobile, messaging, voice, STT/TTS, multimodal transport execution, safe autonomous browser/computer-use, or full browser parity is not claimed by these proof lanes
status: complete on feat/production-reach-hardening-batch-i-v1, intended for the aggregate Batch I PR for #234
root cause addressed:
channel routing only surfaced configured bindings and active adapter ownership, not whether websocket or native delivery were actually reachable at runtime, so operators and delivery audit could not distinguish ready, fallback, and unavailable states
native queued-bundle delivery always reopened ambiently even when every deferred item belonged to the same session, so bundle delivery could break cross-surface thread continuity at the moment the queue drained
the continuity snapshot derived queued bundle thread metadata only from recent intervention lookups and could drop the still-correct queued session_id when an intervention fell outside the recent window
delivered in this slice:
runtime route-health payloads for channel routing and observer continuity, including per-route summaries, selected transport, fallback/unavailable state, and repair hints
runtime-aware delivery selection that uses currently reachable transports rather than static configured order alone
same-thread native bundle continuation when queued items share one session, plus explicit continuation metadata for queued insights and recent interventions
cockpit/settings reach visibility sourced from the same continuity snapshot contract
cd backend && .venv/bin/python -m pytest tests/test_delivery.py -k "falls_back_to_native_when_browser_runtime_is_unavailable or preserves_shared_thread_continuity or deliver_queued_bundle_formats_correctly or no_active_channel_adapters_retains_queue" -q
cd backend && .venv/bin/python -m pytest tests/test_observer_api.py::TestObserverAPI::test_observer_continuity_snapshot tests/test_extensions_api.py::test_channel_routing_defaults_surface_active_adapters -q
cd backend && .venv/bin/python -m pytest tests/test_delivery.py -k "preserves_shared_thread_continuity or partitions_mixed_sessions_into_separate_notifications" tests/test_observer_api.py::TestObserverAPI::test_observer_continuity_recovers_queued_thread_from_intervention_outside_recent_window -q
cd backend && .venv/bin/python -c "import asyncio, json; from src.evals.harness import run_runtime_evals; summary = asyncio.run(run_runtime_evals(['cross_surface_continuity_behavior'])); print(json.dumps({'failed': summary.failed, 'details': summary.results[0].details}, indent=2))"
cd frontend && NODE_OPTIONS=--experimental-require-module npx vitest run src/components/settings/DaemonStatus.test.tsx
cd frontend && NODE_OPTIONS=--experimental-require-module npx vitest run src/components/cockpit/CockpitView.test.tsx
cd frontend && npm run build
review follow-up captured during validation:
the new runtime-health path now reads ws_manager.active_count, so delivery tests that only mocked broadcast results needed explicit active-count fixtures to model a real connected browser runtime
the synthetic “no active adapters” bundle-failure contract now correctly reports inactive+inactive because route health records the actual unavailable transports instead of collapsing everything to the old websocket-disabled code
GitHub CI then exposed the same contract drift inside the eval harness: several reach and delivery scenarios were still using bare websocket or daemon mocks with non-scalar state, so runtime reach inspection could read MagicMock values as transport health instead of explicit connected/disconnected state
that follow-up was fixed in the Batch I branch by normalizing non-scalar websocket and daemon runtime inputs at the channel_routing helper boundary and by pinning explicit active_count fixtures in eval scenarios that are meant to model a live browser websocket path
reviewer: Einstein (019d39cb-8f3b-7e83-8874-9659016bf404)
reviewer findings fixed on the same branch:
mixed-session native bundle delivery was still collapsing unrelated sessions into one ambient notification, so the native bundle path now partitions queued items by session before enqueuing notifications and preserves thread-specific continuation for each session-owned group
queued insights without a stored session_id could still lose thread recovery once the source intervention fell outside the latest-8 recent window, so the continuity snapshot now performs a direct intervention fallback lookup for missing thread context instead of trusting recency alone
the cockpit desktop shell was truncating route-health visibility to the first two routes, so bundle_delivery and later routes are now rendered instead of being silently hidden
full uv run pytest tests/test_eval_harness.py::test_runtime_eval_scenarios_expose_expected_details tests/test_eval_harness.py::test_run_runtime_evals_passes_all_scenarios -q is still failing in this environment for unrelated authenticated-provider scenarios outside Batch I; the Batch I proof therefore uses the scenario-specific cross_surface_continuity_behavior harness run above instead of claiming an unrelated full-harness green result
a subagent review was requested against the uncommitted Batch I reach-hardening diff for bugs, regressions, and hallucinated assumptions; if that review does not return before PR preparation, this branch record should rely on the direct review findings plus the targeted validation above instead of claiming an unreturned clean review
status: complete on feat/production-reach-cross-surface-batch-ap-v1, intended for the aggregate Batch AP PR for #346
root cause addressed:
broader reach continuity still treated route health, imported capability attention, and typed source-adapter degradation as the only actionable non-thread surfaces, so messaging connectors, channel adapters, node adapters, and observer definitions could still disappear behind separate inventories instead of showing up in the same operator recovery loop
observer continuity, cockpit triage, and the desktop shell did not carry a truthful presence-surface-ready versus attention summary, so follow-up-capable surfaces could remain invisible while only repair-oriented reach issues surfaced
the operator timeline and Activity Ledger were still missing the presence-surface counts and recovery kinds needed to keep the widened reach contract visible outside the observer endpoint and cockpit shell
delivered in this slice:
explicit presence-surface inventory in observer continuity across channel adapters, messaging connectors, node adapters, and observer definitions, including ready versus attention summaries, repair hints, and follow-up prompts
cockpit desktop-shell and triage visibility for presence repair plus follow-up flows, including a new top-presence summary line and direct follow-up drafting for ready surfaces
threaded operator timeline and Activity Ledger continuity metadata for presence-surface counts and presence recovery actions
deterministic backend and frontend coverage for the widened continuity contract, plus the pre-batch backend shard fix for the cross_surface_continuity_behavior eval-harness routing seam
cd backend && .venv/bin/python -m pytest tests/test_eval_harness.py -q -k "test_main_lists_available_scenarios"
cd frontend && NODE_OPTIONS=--experimental-require-module npx vitest run src/components/cockpit/seraphPresence.test.ts src/components/cockpit/CockpitView.test.tsx -t "surfaces active triage for approvals, workflows, queued guardian items, and reach failures|deriveSeraphPresenceState"
cd frontend && npm run build
cd docs && npm run build
review follow-up captured during validation:
the first triage proof assumed imported-reach follow-through would still land in the top visible rows after the new presence surfaces were added, but that ordering guarantee was no longer truthful once the widened continuity contract legitimately displaced lower-priority rows
the first desktop-shell pass truncated recovery rows early enough that the new presence_follow_up action could be produced by the backend but never rendered, so the recovery list now exposes enough rows for the new ready-surface follow-up contract to remain actionable
the observer continuity hardening from the pre-batch fix is folded into this slice because partial intervention rows were still failing response validation on missing legacy string fields, and that would have made the new broader continuity contract brittle under older stored payloads
status: complete on feat/reach-cross-surface-batch-az-v1, intended for the aggregate Batch AZ PR for #364
root cause addressed:
broader reach continuity still left packaged browser-provider reach stranded in provider inventory instead of surfacing it through the same observer continuity and recovery contract as other reach surfaces
node adapters only showed up through generic contribution rows, so selected-versus-fallback reach truth, daemon prerequisites, and guarded follow-up posture were missing from the continuity payload
recovery actions only treated messaging and channel surfaces as follow-up-capable, so the widened reach inventory could still stop short of cockpit, timeline, and ledger follow-through control even when the observer payload carried the right reach state
delivered in this slice:
inventory-backed browser-provider presence surfaces with selected-versus-fallback state, provider kind, execution mode, and network or daemon prerequisite truth
inventory-backed node-adapter presence surfaces with adapter kind, runtime-state follow-up eligibility, and daemon or network prerequisite truth
the same presence-follow-up recovery contract widened to browser-provider and node-adapter surfaces, so observer continuity can drive cockpit triage, the desktop shell, the threaded operator timeline, and the Activity Ledger from the same richer reach inventory
deterministic observer and eval-harness proof for the widened reach inventory contract
cd backend && .venv/bin/python -m pytest tests/test_observer_api.py -q -k "presence_surface_payload or continuity_surfaces_imported_reach_and_source_adapter_attention"
cd backend && .venv/bin/python -m pytest tests/test_eval_harness.py -q -k "cross_surface_continuity_behavior or test_main_lists_available_scenarios"
cd docs && npm run build -- --out-dir /tmp/seraph-docs-build-az
git diff --check
review follow-up captured during validation:
verified subagent review found that removing browser providers and node adapters from the generic presence path made planned, overridden, and invalid packaged reach disappear whenever inventory rows were absent or sanitized, so observer continuity now preserves those generic fallback attention surfaces and lets richer inventory rows overwrite them only when they exist
verified subagent review also found that the first browser-provider follow-up proof asserted an impossible packaged ready state, so the continuity contract now treats selected staged_local_fallback providers as the truthful guarded follow-up path and pins that fallback-aware behavior in tests and the eval harness
the first inventory-backed observer pass also imported default_manifest_roots_for_workspace and settings from the wrong modules and referenced package_label_by_id before constructing it locally, so those direct validation bugs were fixed on the same branch before the review follow-up landed
status: complete on feat/reach-cross-surface-batch-az-v1, intended for the aggregate Batch AZ PR for #364
root cause addressed:
broader reach continuity still left packaged browser providers out of the explicit presence-surface inventory, so selected browser reach and degraded remote-browser fallback could remain invisible in the same recovery loop that already carried messaging and adapter continuity
node adapters were still represented only through the generic presence surface contract, so companion or device reach could not expose adapter-specific follow-up or daemon/network prerequisite truth
the reach eval proof still only pinned the earlier messaging/adapter presence contract, so browser-provider and node-adapter continuity could drift without a deterministic observer or eval-harness seam catching it
delivered in this slice:
inventory-backed browser-provider presence surfaces in observer continuity, including selected versus fallback state, execution mode, provider kind, and network/daemon prerequisite truth
inventory-backed node-adapter presence surfaces in observer continuity, including adapter kind, staged-link readiness, and network/daemon prerequisite truth
the same broader reach follow-up and repair contract now extends through browser-provider and node-adapter surfaces instead of stopping at generic messaging or adapter inventory
deterministic backend proof for the widened observer continuity payload and the cross-surface continuity eval details that consume it
cd backend && .venv/bin/python -m pytest tests/test_observer_api.py -q -k "presence_surface_payload or continuity_surfaces_imported_reach_and_source_adapter_attention"
cd backend && .venv/bin/python -m pytest tests/test_eval_harness.py -q -k "cross_surface_continuity_behavior or test_main_lists_available_scenarios"
cd docs && npm run build -- --no-minify
git diff --check
review follow-up captured during validation:
the first inventory-backed observer pass referenced a package-label lookup that was never built inside _observer_presence_surface_payload(), so browser-provider and node-adapter continuity would have failed before returning any payload; the branch now builds that lookup directly from list_extensions() before constructing inventory-backed surfaces
a subagent review was requested twice for this Batch AZ diff, but both review passes timed out without returning findings before PR preparation; this branch record therefore relies on the direct validation findings above instead of claiming an unreturned clean review