Purpose: One short page for agents and humans: what is already shipped or closed, with pointers to canonical roadmaps and backlog [DONE] rows. Living master table: master-roadmap-and-refactor-plan-2026-04-25.md §2 + §0; structural debt: ../backlog/refactor-backend.md, ../backlog/refactor-frontend.md. Authoritative gate numbers: eval/results/benchmark-trust-baseline.json (not §10 narrative tables).
Last rolled up: 2026-05-04.
Staleness note (2026-05-15): this is a historical rollup snapshot, not a live queue. For active priorities use README.md and live plans linked there.
| Track | Done / partial (high level) |
|---|---|
| A Ingest async | Waves U/V/W delivered — _archive/ingestion-async-pipeline-roadmap-2026-04-25.md. |
| B LangGraph | Y1–Y4 done; Y5/Y6 open — langgraph-migration-plan-2026-04-25.md. |
| C Phoenix | X1 / X2 done; X3 partial (producer inject + worker extract, 2026-04-27); Wave X plan CLOSED — full text _archive/phoenix-tracing-coverage-2026-04-25.md, stub phoenix-tracing-coverage-2026-04-25.md, evidence phoenix-closeout-evidence-2026-04-27.md. |
| D Benchmarks / gold | Corpus Gold Pack v1 Phase 0–6 + BT0; BT1, BT5, BT3 pilot (Wave 6); M and gold layers per master §4.D — trust plan ontology-benchmarks-trust-audit-2026-04-25.md. |
| E Graph UX | GR1, GR3 (caveats), GR5 API slice; GR2 backend done / UI via GR6 — graph-readability-followup-2026-04-25.md. |
| F Workspace UX | I/J/K/L1–L2, WX1 (2026-04-26), WX5 minimal (2026-04-27) — workspace-ux-redesign-2026-04-25.md. |
| RX / LX | RX1 partial; LX1 settings + alias, LX2 SSE stub + schema (2026-04-27) — reader-ux-and-translation-roadmap-2026-04-25.md. |
Status: Phase 0–6.E complete (gold + dual/triple validation). Detail log (archived): _archive/corpus-gold-pack-v1-phase-log-2026-04-25.md. Fixture layout + layers §1–4: still in corpus-gold-pack-v1-2026-04-25.md.
Numbers (headline): 71 packs total → 35 promoted (33 llm_dual_validated + 2 llm_triple_validated); 36 high-priority for human review; 9 gold layers + catalog/relations. BT0 narrative — ontology-benchmarks-trust-audit-2026-04-25.md §5 BT0.
Plan: agent-chat-frontend-ui-plan-2026-04-26.md. System roadmap (slim): agent-runtime-tools-context-roadmap-2026-05-04.md — упрощённая архитектура; rule-based tool_search v1 shipped (science_graphrag/agent/tool_search.py); compaction / session memory remain phased; полный старый текст в _archive/chat-agent-system-roadmap-full-2026-04-26.md.
| Phase | Status |
|---|---|
| UI-1 | DONE — shimmer turn shell (ShimmerLabel, AgentRunHeader, AgentLiveStatus, ChatMessageThread). |
| UI-2 | DONE — AskAnswerPanel run chrome + collapsed inspector. |
| UI-3 | DONE — subagent rail (AgentSubagentRail / AgentSpecialistRunStack, buildSpecialistStreamGroups / shouldShowSubagentRail in agentRunViewModel.js). |
| UI-4 | DONE — typed block chrome (TYPED_BLOCK_OUTER_SX / ChatTypedBlocks.jsx). |
| UI-5 | PARTIAL — product_step / synthesis SSE + view-model headline/dedup/post-run line; дальнейшие типы событий — по необходимости (см. UI plan §11). |
Tests (§12.1): stream parse / subagent cards / shimmer / warnings — DONE; full answer + all typed blocks RTL — partial; manual SSE — OPEN (чеклист: agent-chat-frontend-verification-gaps-next-wave.md).
Evidence IA (2026-04-27): DONE — primary sidebar no longer links to /evidence; canonical inspection URLs use buildStandaloneEvidencePath; citations and Reader trace banner open /evidence with optional workspace_id; legacy unmounted tabs OverviewTab / GraphTab / EvidenceTab removed. See recorded choice in agent-chat-frontend-ui-plan-2026-04-26.md and Phase 5 notes in ../specs/ui-ux-master-plan.md.
P0 (truthful graph for methods mode + citations / full 1-hop contract): DONE (2026-04-27) — full RCA _archive/workspace-graph-methods-citations-root-cause-2026-04-27.md, stub workspace-graph-methods-citations-root-cause-2026-04-27.md.
DONE (shipped 2026-04-26) — _archive/wave5-bt6-quote-tolerance-2026-04-26.md. Live trust_signal / gold realism follow-ups remain in backlog.
Rounds 1–5 closed — _archive/completed-rounds-2026-04-25.md. Round 6+ partial / in progress — see master-roadmap-and-refactor-plan-2026-04-25.md §7–§10.
GR1–GR3 (with caveats) and migration of GR4→GR9 — header in _archive/graph-ux-aggregation-roadmap-2026-04-25.md; active follow-up — graph-readability-followup-2026-04-25.md.
No “shipped checklist” in ingestion-llm-architecture-and-instructor-standardization-2026-04-27.md — planning doc only; execution rows land in backlog + PRs.
docs/analysis/README.md: single planning hub — weekly pointers (trust-audit, backlog, snapshot,benchmark-trust-baseline.json); §10 of master documented as historical log, not live queue; full inventory of root.mdfiles; Closed plans with stubs + copies under_archive/(Phoenix Wave X, graph work vs workspace DRY Phases 0–5, HashRouter remediation, workspace graph methods/citations RCA).- Graph / nav closure: DRY plan Phases 0–5 and hash-router remediation DONE — full text under
_archive/graph-work-vs-workspace-unification-dry-plan-2026-04-28.md,_archive/graph-navigation-hash-router-remediation-plan-2026-04-28.md. - Habr / claims narrative: measurement spine and pinned artifacts —
habr-article-narrative-and-measurement-plan-2026-07.md; gold-v2 wave rows andeval/results/habr-window-2026-07-*(+ Appendix A in../benchmarks/ontology-claims-benchmark-v1.md). - Agent
tool_search: rule-based shortlist + tests (science_graphrag/agent/tool_search.py,tests/test_tool_search.py); eval hooks documented in benchmark specs — not a future-only track.