{"quests":[{"name":"Agent Ecosystem","description":"Background agent fleet management — agent registry, health monitoring, task routing, and auto-scaling. Define agent roles (researcher, debater, market-maker, auditor), spawn/retire agents based on workload, and maintain agent identity across sessions. Goal: a self-managing ecosystem of specialized agents.","layer":"Forge","quest_priority":96,"quest_status":"active","done":23,"running":0,"open":3,"total":31},{"name":"Causal Mechanism & Upstream Targets","description":"LLM-augmented causal reasoning for neurodegeneration. Build weighted causal evidence graphs, rank candidate mechanisms by causal strength, and select earliest-/most-upstream druggable targets. Output: research plans agents can execute. Integrates DoWhy/PyWhy-style causal effect estimation with the SciDEX KG, debate transcripts, and Mendelian-randomization evidence. References Kiciman & Sharma 2023 on LLM causal reasoning frontier.","layer":"Cross-cutting","quest_priority":96,"quest_status":"active","done":27,"running":0,"open":0,"total":27},{"name":"Epistemic Rigor","description":"Evolve SciDEX toward rigorous scientific epistemology: 6-tier epistemic classification (T0 axiom → T5 contested) with asymmetric promotion/demotion (Gödel Machine principle), adversarial Falsifier persona as 5th debate round, proof-gated KG updates with contradiction checking against established knowledge, GFlowNet-inspired diverse hypothesis portfolio maintenance (anti-monoculture rules), pre-registration of predictions before analysis, replication status tracking (unreplicated/replicated/failed_replication), and calibration-adjusted believability weights. Every claim must be traceable to ground truth.","layer":"Cross-cutting","quest_priority":95,"quest_status":"active","done":23,"running":0,"open":0,"total":23},{"name":"Funding-Linked Prioritization","description":"Connect landscape/gap analyses to Exchange funding allocator + artifact prioritization. Each landscape thinness signal raises a funding bid; each high-quality gap auto-becomes a mission with token bounty; ROI of funded missions feeds back into next-cycle priority. Closes the 'rank → fund → execute → re-rank' loop end-to-end.","layer":"Exchange","quest_priority":95,"quest_status":"active","done":9,"running":0,"open":0,"total":9},{"name":"Open Questions as Ranked Artifacts","description":"First-class open_question artifact_type with per-field ranked leaderboards. Auto-populates from: existing wiki page 'Open Questions' / 'Unanswered' sections (18,447 wiki pages to mine), high-quality gaps, falsifiable hypothesis predictions, debate dissent points. Ranking via Elo tournaments scoped per field_tag (importance × tractability × potential_impact). Per-field views: /questions/aging, /questions/connectomics, /questions/synthetic-biology, etc. Each question is debatable, discussable, citable; can be promoted to an experiment_proposal or analysis_proposal. Pairs with Q-LIVE so per-field ranked views are live dashboards, not static pages. Continual: questions decay if stale, get re-ranked on new evidence, are auto-archived when resolved.","layer":"Atlas","quest_priority":95,"quest_status":"active","done":23,"running":0,"open":0,"total":23},{"name":"Search","description":"Universal search across all SciDEX content with autocomplete","layer":"Cross-cutting","quest_priority":95,"quest_status":"active","done":24,"running":0,"open":1,"total":209},{"name":"Universal Artifact Discussions","description":"Wikipedia-style talk-page on EVERY artifact. The artifact_comments table already exists (threaded, comment_type, reactions, depth) but has 0 rows. This quest activates it across the platform: thread UI on every artifact detail page (hypothesis, wiki, notebook, dataset, dashboard, gap, proposal), API for listing/posting/reacting, mention-based cross-artifact linking (@uuid in comment → artifact_link), watchlists + notifications. Distinct from q-artifact-debates (structured multi-persona debates) — this is lightweight, perpetual, asynchronous discussion. Pairs with Q-PERC for percolation of comments into actions, edits, and cross-links. Long-running quest — every kind needs to gain a thread and every flow needs to honor mentions.","layer":"Cross-cutting","quest_priority":95,"quest_status":"active","done":14,"running":0,"open":0,"total":14},{"name":"Exchange","description":"Hypothesis prediction market — score, track, and compare hypotheses","layer":"Exchange","quest_priority":95,"quest_status":"active","done":342,"running":0,"open":5,"total":1026},{"name":"Open Debates","description":"Rich multi-agent debates with open participation — any agent can join a debate session, contribute evidence, challenge claims, and vote. Includes debate enrollment protocol, structured rounds with evidence requirements, spectator mode, and debate outcome synthesis. Goal: move beyond fixed persona debates to dynamic, ecosystem-wide deliberation.","layer":"Agora","quest_priority":94,"quest_status":"active","done":22,"running":0,"open":4,"total":27},{"name":"Continuous Proposal Generation","description":"Generate actionable items from SciDEX continuously: analyses, experiments, code, grants. New artifact_types: analysis_proposal, code_proposal, experiment_proposal, funding_proposal. Each is rankable (Elo scoped per kind, plus composite), discussable (Q-DSC), versioned (artifacts table), traceable (processing_steps + provenance_chain). Generators: from-hypothesis, from-gap, from-discussion, from-collider-session, from-pattern (cross-paper signals), from-debate-dissent. Conversion paths: code_proposal + Senate approval → Orchestra task; experiment_proposal → IRB-ready protocol; analysis_proposal → Jupyter notebook stub; funding_proposal → grant draft. Continual: every gap should produce 1-3 proposals; every top-N hypothesis should produce ≥1 experiment_proposal; every active discussion thread can spawn proposals. The system's externally visible value is this proposal stream.","layer":"Cross-cutting","quest_priority":94,"quest_status":"active","done":13,"running":0,"open":0,"total":13},{"name":"Economics","description":"Resource economics — token/cost tracking, belief evolution, impact scoring. Extends to quadratic staking (agents stake reputation quadratically for high-confidence assertions), calibration tracking (per-persona hit rate on predictions vs outcomes), GFlowNet-inspired diversity bonuses (reward-proportional resource allocation preserving competing hypotheses), and contested claim prediction markets.","layer":"Cross-cutting","quest_priority":94,"quest_status":"active","done":15,"running":0,"open":1,"total":200},{"name":"Agora","description":"Multi-agent debate engine — any agent can participate in scientific debates using real evidence. Open enrollment debate protocol, structured argumentation, evidence citation, and cross-agent deliberation.","layer":"Agora","quest_priority":94,"quest_status":"active","done":256,"running":1,"open":6,"total":287},{"name":"Experiment Extraction","description":"Extract structured experimental findings from papers with full lineage — methods, results, statistics, context","layer":"Atlas","quest_priority":93,"quest_status":"active","done":19,"running":0,"open":1,"total":20},{"name":"Percolation Engine","description":"Closes the loop: discussion comments and proposals percolate into concrete actions, edits, cross-links, and new artifacts — everything tracked end-to-end. Components: (1) comment classifier (LLM tags posts as proposal/action/evidence/dispute/question/ack); (2) action emitter — when N agents reach consensus on action-tagged thread, spawn Orchestra task with comment_id provenance; (3) edit emitter — when consensus on edit-proposal, bump artifact version applying the diff and credit contributors; (4) cross-link emitter — when comments link ≥3 artifacts, propose typed artifact_link; (5) proposal emitter — when discussion converges on a proposal shape, spawn the right artifact (analysis_proposal etc.); (6) percolation dashboard — daily flow rates of post→task→artifact-version. Audit + provenance preserved: every action references the originating comment_id and contributors, restoring full reproducibility from the discussion all the way to the merged change.","layer":"Senate","quest_priority":93,"quest_status":"active","done":18,"running":0,"open":0,"total":18},{"name":"Autonomous Engines","description":"Orchestra-managed recurring scripts: debate engine, link checker, orphan checker, convergence monitor, visual regression, pubmed pipeline, metrics, health checks","layer":"Senate","quest_priority":93,"quest_status":"active","done":14,"running":0,"open":4,"total":18},{"name":"Artifact Governance & Lifecycle Management","description":"Establish a unified artifact governance system across SciDEX. Every first-class entity (gap, hypothesis, analysis, paper, dataset, wiki page, debate, squad finding) gets a universal artifact identity, full history tracking, merge/split/archive operations with decision audit trails, and content ownership by agents or squads.\n\nEXISTING INFRASTRUCTURE (to build on, not replace):\n- dedup_recommendations table: 297 pending merge candidates, zero executed\n- artifact_transforms table: 15 records tracking versions/lifecycle/deprecation\n- API endpoints: /api/dedup/{recommendations,approve,reject,scan}, /api/artifact-transforms\n- Lifecycle states: hypotheses (7 states), gaps (6 states), datasets (versioned via git)\n- governance_config table: 3 entries (ALLOCATION_THRESHOLD, REQUIRED_SIGNATURES, AUTHORIZED_SIGNERS)\n- edit_history table: schema exists, zero rows (never wired up)\n- quality_gate_results table: schema exists, zero rows\n\nGAP ANALYSIS (what's missing):\n1. No universal artifact registry — each type has its own ID scheme\n2. edit_history never wired up — no who-changed-what audit trail\n3. 297 dedup recommendations pending forever — no execution pipeline\n4. Lifecycle enforcement is manual/absent — only 1 of 337 hypotheses archived\n5. No content ownership model — artifacts are unowned\n6. No governance decision tracking — decisions happen but aren't recorded\n7. No automated garden maintenance — stale gaps, redundant hypotheses accumulate\n\nDESIGN PRINCIPLES:\n- Clean system APIs: every operation available via /api/ endpoints\n- Documented governance processes adopted by ALL subsystems\n- Agent-compatible: agents and squads can propose, vote, and execute decisions\n- Incremental: builds on existing tables, doesn't require big-bang migration\n- Auditable: every state change has a who/when/why trail","layer":"Senate","quest_priority":93,"quest_status":"active","done":28,"running":0,"open":0,"total":28},{"name":"Resource Intelligence","description":"Value-based resource allocation — continually evaluate the relative value of quests, tasks, \nhypotheses, gaps, and artifacts, then allocate LLM tokens, CPU, and disk proportionally.\n\nThis quest is the DECISION layer between:\n  - Economics (pricing signals, token economy) → tells us what things are WORTH\n  - Resource Governance (caps, quotas, enforcement) → tells us what we CAN spend\n  - Resource Intelligence (THIS) → decides what we SHOULD spend on\n\nCore loop:\n  1. Score every open task/quest by expected-value-per-token:\n     EV = (importance × tractability × downstream_impact) / estimated_cost\n  2. Rank by EV descending → priority queue\n  3. Allocate budget: top-k items get resources proportional to EV rank\n  4. After completion, measure actual impact → update EV model (feedback loop)\n\nSignals feeding the scorer:\n  - Elo tournament ratings (hypothesis quality signal)\n  - LMSR market prices (belief-weighted value)\n  - Gap value_if_resolved scores\n  - Quest priority × task count\n  - Historical cost-per-completion from resource_usage + cost_ledger\n  - Freshness: penalize stale items, boost newly-identified opportunities\n\nResources tracked:\n  - LLM tokens (by provider: max, codex, bedrock)\n  - CPU seconds (cgroup tracking via cgroup_isolation.py)\n  - Disk (artifact storage)\n  - Agent slots (concurrent execution capacity)\n\nConnects to:\n  - resource_allocations table (currently empty → needs populating)\n  - quest_resource_analyzer.py (correlates completion with consumption)\n  - monthly_resource_adjustment.py (periodic rebalancing)\n  - token_ledger.py (virtual economy)\n  - cost_ledger in orchestra.db (real spend)\n\nSpec: docs/planning/specs/q-resource-intelligence_spec.md","layer":"Senate","quest_priority":92,"quest_status":"active","done":8,"running":0,"open":1,"total":9},{"name":"Crypto Wallets","description":"Crypto wallet integration for agents — each agent gets a wallet for receiving and allocating funds. Supports token-based capital allocation, on-chain audit trails, multi-sig governance for large allocations, and integration with Exchange market mechanisms. Foundation for decentralized agent-driven research funding.","layer":"Exchange","quest_priority":92,"quest_status":"active","done":13,"running":0,"open":0,"total":13},{"name":"Artifact Quality Markets","description":"Market pricing for all artifact types — participant agents, lifecycle governance, reputation staking, debate feedback","layer":"Exchange","quest_priority":92,"quest_status":"active","done":13,"running":0,"open":1,"total":14},{"name":"Real Data Pipeline","description":"Ensure all analyses use REAL Allen Institute datasets (SEA-AD, ABC Atlas, Allen Brain Cell Atlas) — not simulated/hallucinated data. Build data ingestion pipelines that download, cache, and validate real datasets. Analyses must cite specific dataset versions and cell counts. Forge tools must be invoked during the analysis loop with real inputs, not just registered as provenance markers.","layer":"Forge","quest_priority":92,"quest_status":"active","done":16,"running":0,"open":0,"total":16},{"name":"Live Dashboard Artifact Framework","description":"Dashboard artifact_type already exists (6 test rows with rendered_html, last_render_data, last_render_errors fields). This quest builds it into a first-class, sustainable framework: declarative view_spec_json with safe-query DSL (read-only parameterized PG queries), render path with caching + error capture, POST /api/dashboard/{id}/snapshot creates immutable child artifact (citation-grade), wiki pages can embed live dashboard widgets. Seed library: top-hypotheses-by-field, open-questions-by-field, recent-proposals, gap-flow, debate-leaderboard, evidence-density-map, pipeline-flow (landscape→gap→debate→hypothesis→proposal). Continual: every operator question 'what's the state of X?' becomes a dashboard. Static wiki pages stay for canonical knowledge; dashboards are the live counterpart.","layer":"Atlas","quest_priority":92,"quest_status":"active","done":17,"running":0,"open":0,"total":17},{"name":"Artifact Debates","description":"Make any artifact debatable — evidence accumulation through debates, usage, and citations","layer":"Agora","quest_priority":91,"quest_status":"active","done":12,"running":0,"open":0,"total":12},{"name":"Work Governance","description":"Comprehensive governance framework for tracking all work added to SciDEX — who did what, when, why, and at what cost. Includes commit-level provenance, agent accountability scores, automated quality checks, approval workflows, contradiction detection before KG modifications, rollback capability via kg_modifications log, artifact transform audit trail (merge/split/deprecate/supersede operations), supersession tracking, and dashboards showing contribution patterns.","layer":"Senate","quest_priority":90,"quest_status":"active","done":6,"running":0,"open":2,"total":9},{"name":"Analysis Sandboxing","description":"Run all analyses in isolated environments (containers, cgroups, or sandboxed subprocesses) so a runaway analysis cannot corrupt the database, filesystem, or other running services. Each analysis gets its own temp directory, restricted filesystem access, and resource budget. Design for future migration to external container runtimes (Docker, AWS Batch, K8s) while running locally today.","layer":"Forge","quest_priority":90,"quest_status":"active","done":13,"running":0,"open":0,"total":13},{"name":"Multi-Source Literature Search","description":"Expand SciDEX's literature search from PubMed-centric to multi-source, covering Paperclip MCP (8M+ biomedical papers across arXiv/bioRxiv/PMC/OpenAlex/OSF), Semantic Scholar (citation networks + influence metrics), OpenAlex (already partial), and CrossRef (already partial). SciDEX already has a PaperCorpus abstraction layer with adapter classes — this quest extends it with new adapters, strengthens the citation/reference model beyond PMID as primary key, refactors legacy PubMed-only pipelines to use the generic layer, and integrates Paperclip MCP for agent-native deep-dive capabilities (search, grep, map, ask-image, sql). Goal: any SciDEX component that searches literature or cites papers goes through a unified multi-source interface.","layer":"Atlas","quest_priority":90,"quest_status":"active","done":13,"running":0,"open":0,"total":13},{"name":"Tool Playground","description":"Interactive tool interfaces with live API calls, structured results, external links","layer":"Forge","quest_priority":88,"quest_status":"active","done":0,"running":0,"open":0,"total":1},{"name":"Schema Governance","description":"Agent-driven schema evolution through governance (propose, debate, vote, migrate). Linked to artifact_transforms for tracking schema change provenance. All schema changes recorded as artifact lifecycle events.","layer":"Senate","quest_priority":88,"quest_status":"active","done":7,"running":0,"open":0,"total":7},{"name":"Code Health","description":"Continuous code quality review without removing functionality. Archive dead code to cold storage (never delete), organize file sprawl into packages, consolidate duplicate functions, break apart api.py. Connected to artifact lifecycle governance — archival uses cold_store mechanism.","layer":"Senate","quest_priority":88,"quest_status":"active","done":15,"running":0,"open":1,"total":16},{"name":"Resource Governance","description":"Enforce hard resource limits on the VM — memory caps per analysis, CPU quotas, concurrent analysis limits, cumulative daily cost budgets, and auto-kill for runaway processes. Track and display resource usage in real-time. Prevent the VM from becoming unresponsive due to analysis load. Architecture should support future offloading to external compute (AWS Batch, Lambda, K8s) via a pluggable executor interface.\n\n[2026-04-09] See also q-resource-intelligence for the allocation DECISION layer (this quest handles enforcement/limits, that one handles value-based allocation).","layer":"Senate","quest_priority":88,"quest_status":"active","done":9,"running":0,"open":0,"total":9},{"name":"Senate","description":"Governance & quality gates — agent performance tracking, work audit trails, convergence monitoring, quality control. Includes epistemic tier enforcement (T0-T5), adversarial falsification via red-team Falsifier persona, proof-gated KG updates with contradiction checking, and epistemic health dashboards. Implements the Epistemic Ratchet: adding knowledge is easy, degrading established knowledge is hard.","layer":"Senate","quest_priority":88,"quest_status":"active","done":415,"running":1,"open":23,"total":810},{"name":"Gap Factory","description":"Automated knowledge gap generation, decomposition, and value-scoring pipeline.\n\nCore principles:\n- Gaps are almost never \"resolved\" — they represent open questions that require extraordinary convergent evidence (5+ analyses, 2+ debates, composite score ≥0.85) to close.\n- Gaps have VALUE proportional to: importance × log(1 + dependent_hypotheses) × tractability.\n- Gaps decompose into subproblems (gap_subproblems table) and form dependency networks (gap_dependencies: blocks / informs / subsumes / requires / contradicts).\n- New gaps are identified by: debate transcript analysis, KG contradiction detection, paper frontier scanning, hypothesis-gap mismatch analysis.\n\nStatus model (graduated evidence thresholds):\n  open → investigating → partially_addressed → substantially_addressed → resolved\n\nInfrastructure delivered (2026-04-06):\n- Migration 054: gap_dependencies, gap_subproblems, value_if_resolved column, resolution_criteria\n- update_gap_statuses() with graduated thresholds\n- Dashboard shows \"Active Gaps\" (non-resolved) instead of \"Open Gaps\"\n\nOpen workstreams:\n1. Gap identification: scan new papers/debates for unresolved questions\n2. Gap decomposition: LLM-driven subproblem extraction from gap descriptions\n3. Gap dependency mapping: connect gaps to hypotheses and other gaps\n4. Gap value scoring: count downstream impact on hypothesis portfolios\n5. Gap tournaments: Elo-rank gaps by \"most important to solve\" via LLM judge\n6. Gap-hypothesis matching: surface which hypotheses address which gaps","layer":"Atlas","quest_priority":85,"quest_status":"active","done":10,"running":0,"open":2,"total":12},{"name":"Forge","description":"Scientific execution engine — tool registry, background agent orchestration, agent spawning, and tool-augmented analysis. Manages the ecosystem of background agents that perform research tasks.","layer":"Forge","quest_priority":80,"quest_status":"active","done":183,"running":0,"open":10,"total":564},{"name":"Atlas","description":"Living knowledge graph + world model — multi-representation scientific knowledge. Includes automated gap detection pipeline, gap prioritization, and continuous knowledge frontier expansion.","layer":"Atlas","quest_priority":80,"quest_status":"active","done":324,"running":1,"open":18,"total":712},{"name":"Capital Markets","description":"Virtual token economy with bidding, portfolios, settlement, and capital-weighted resource allocation — simulating blockchain","layer":"Exchange","quest_priority":77,"quest_status":"active","done":17,"running":1,"open":6,"total":24},{"name":"Market Participants","description":"Market participant agents that evaluate and allocate capital toward ideas, challenges, gaps, proposals, and experiments. Agents bid on hypothesis quality, fund gap investigations, and stake on debate outcomes. Includes portfolio management, risk assessment, and ROI tracking. Capital flows signal which research directions the ecosystem values most.","layer":"Exchange","quest_priority":75,"quest_status":"active","done":7,"running":0,"open":4,"total":11},{"name":"Artifact Lifecycle","description":"Git-like artifact lifecycle governance: lifecycle state machine (active→deprecated→cold→superseded), supersession chains with forward reference resolution, dependency-aware versioning (new version = new commit), deduplication agent scanning hypotheses/wiki/artifacts for duplicates, merge recommendation review workflow, transform audit log, and governance dashboard at /senate/artifact-governance. Core principle: never delete, always version, full provenance.","layer":"Senate","quest_priority":75,"quest_status":"active","done":10,"running":0,"open":0,"total":10},{"name":"Wiki","description":"NeuroWiki world model integration — rich wiki pages with mermaid, infoboxes, cross-links","layer":"Atlas","quest_priority":75,"quest_status":"active","done":23,"running":0,"open":2,"total":394},{"name":"Evolutionary Arenas","description":"Elo tournaments × LMSR markets × evolutionary operators for scientific artifacts. Pairwise LLM-judged matches produce quality signals that fuse with market prices; top-ranked artifacts spawn mutate/crossover/refine variants; adaptive loops dig deeper on winners. Bridges Bradley-Terry ≡ Elo ≡ LMSR via shared functional form. Spec: docs/planning/specs/q-evolutionary-arenas_spec.md","layer":"Cross-cutting","quest_priority":73,"quest_status":"active","done":9,"running":0,"open":1,"total":10},{"name":"Adversarial Science","description":"Red-team validation for scientific claims. Falsifier persona as 5th debate round searches for contradictions, counter-evidence, and logical flaws. Falsification scores gate hypothesis promotion — score > 0.7 blocks promotion above T3 (provisional) without human review. Draws on AI2 Theorizer adversarial approach and Popperian falsificationism.","layer":"Senate","quest_priority":70,"quest_status":"active","done":9,"running":0,"open":0,"total":9},{"name":"Resources","description":"Resource tracking — Bedrock costs, API call logging, budget monitoring","layer":"Cross-cutting","quest_priority":70,"quest_status":"active","done":1,"running":0,"open":0,"total":1},{"name":"Artifacts","description":"Rich artifacts & computational notebooks. Artifact lifecycle governance: git-like versioning (version_number, parent_version_id, changelog), lifecycle states (active/deprecated/cold/superseded), supersession chains with forward reference resolution, dependency tracking, never-delete policy. Deduplication agent scans for duplicates and produces merge recommendations for governance review.","layer":"Cross-cutting","quest_priority":68,"quest_status":"active","done":102,"running":0,"open":4,"total":474},{"name":"UI","description":"UI/Infrastructure — consistent styling, responsive design, loading states","layer":"Cross-cutting","quest_priority":66,"quest_status":"active","done":208,"running":0,"open":4,"total":949},{"name":"AI Tools Landscape","description":"Living catalog of AI-for-science tools, platforms, agents, and benchmarks. Uses wiki entity_type='ai_tool' with structured frontmatter (category, specializations, pricing, benchmarks, open-source status). Browsing page at /forge/landscape with filters by category, specialization, pricing. Seeded with 23 entries from inspirations page. Kept current via agent scanning + staleness badges. Spec: docs/planning/specs/q-ai-tools-landscape_spec.md","layer":"Forge","quest_priority":64,"quest_status":"active","done":5,"running":0,"open":1,"total":9},{"name":"Content Quality Sweep","description":"HIGHEST PRIORITY: Systematic review of all top-level pages and content for quality issues. The system has accumulated junk: 7,824 broken figure artifacts, duplicate test hypotheses, placeholder notebooks, and low-quality auto-generated content. Quality gates are post-hoc only. This quest gates ALL future content generation behind quality verification.","layer":"Senate","quest_priority":62,"quest_status":"active","done":11,"running":0,"open":4,"total":15},{"name":"Hypothesis Diversity","description":"GFlowNet-inspired diverse hypothesis portfolio maintenance. Instead of collapsing to top-scoring hypotheses, maintain competing explanations across different mechanistic families. Anti-monoculture rule: if >70% of hypotheses for a gap share the same mechanism, trigger exploration of alternatives. Diversity score measures how many distinct mechanistic explanations are represented. Resource allocation proportional to composite_score × diversity_bonus.","layer":"Exchange","quest_priority":60,"quest_status":"active","done":10,"running":0,"open":0,"total":10},{"name":"Site Analytics","description":"Track visitor engagement — pageviews, sessions, bounce rate, referrers, content type breakdown, search queries. Lightweight middleware records every request. Dashboard at /senate/analytics with trend charts, top pages, and improvement signals. Privacy-preserving (session = IP+UA hash, no PII). Spec: docs/planning/specs/q-site-analytics_spec.md","layer":"Senate","quest_priority":55,"quest_status":"active","done":4,"running":0,"open":0,"total":4},{"name":"Pre-Registration & Replication","description":"Pre-registration: before each analysis, agent commits expected findings (logged immutably). Large deviations between prediction and finding trigger scrutiny flag. Replication tagging: claims from single extraction method get replication_status=unreplicated. Independently confirmed claims get replicated. Contradicted get failed_replication. Unreplicated high-impact claims get priority for re-investigation.","layer":"Cross-cutting","quest_priority":55,"quest_status":"active","done":6,"running":0,"open":0,"total":6},{"name":"Visual Artifacts","description":"AI-generated visual artifacts — infographics, hypothesis cards, entity portraits, protocol illustrations. Uses MiniMax image-01 and GLM CogView-4 with quality evaluation via vision models.","layer":"Forge","quest_priority":55,"quest_status":"active","done":5,"running":0,"open":0,"total":5},{"name":"LLM Routing & Provider Management","description":"Route tasks to appropriate LLMs based on characteristics — MiniMax for throughput, Claude for safety","layer":"","quest_priority":50,"quest_status":"active","done":7,"running":0,"open":0,"total":7},{"name":"Demo","description":"Demo showcase & end-to-end flows — polished demos exercising all five layers","layer":"Cross-cutting","quest_priority":50,"quest_status":"active","done":290,"running":0,"open":4,"total":607},{"name":"Deep Site Audit: Logic, Rigor & Credibility","description":"Comprehensive hackathon-judge review of SciDEX. 10 sequential agents each audit the site walkthrough, identify issues with agentic loops, debate quality, data coherence, UX — then implement fixes. Focus: debate pipeline rigor, hypothesis scoring credibility, gap-to-investigation flow, empty states, market concept explanations, narrative coherence.","layer":"Agora","quest_priority":10,"quest_status":"active","done":1,"running":0,"open":0,"total":1},{"name":"Other","description":"Tasks without a quest assignment","layer":"","quest_priority":0,"quest_status":"active","done":1487,"running":0,"open":8,"total":2253}],"running":[{"id":"80ffb77b-8391-493c-8644-37086c8e2e3c","title":"[Senate] CI: Ambitious quest task generator — xhigh-effort LLM-driven strategic task creation","started":"2026-04-13T09:59:27.424929+00:00","updated":"2026-04-29T06:27:23.999277+00:00","priority":99},{"id":"1544b567-7581-467f-9a9e-1605913998b9","title":"[Atlas] Drug target therapeutic recommendations — generate actionable recs for 91 tier-1 neurodegeneration targets","started":"2026-04-29T04:48:02.286046+00:00","updated":"2026-04-29T06:27:11.761481+00:00","priority":96},{"id":"21ef1f77-5b0c-4d00-81af-ddfcf08ce9f9","title":"[Exchange] Token demurrage sweep (driver #17)","started":"2026-04-11T06:13:36.279388+00:00","updated":"2026-04-29T06:27:02.214651+00:00","priority":91},{"id":"2c4b95b0-39d3-4ae8-b067-f9ae1241aec3","title":"[Agora] Falsifiable prediction evaluation pipeline — auto-score 3,741 pending predictions against literature","started":"2026-04-29T03:16:41.964066+00:00","updated":"2026-04-29T06:25:58.017085+00:00","priority":92}],"recent":[{"id":"3420512b-ab36-4f01-bdee-5a4508fda2e6","title":"snap-getnext-1","completed":"2026-04-29T05:57:15.267658+00:00","priority":50},{"id":"e5a5e66c-acac-4815-bf3d-8d3ce84feaad","title":"snap-list-1","completed":"2026-04-29T05:57:15.023751+00:00","priority":50},{"id":"aa1ae536-d1bf-44f3-9e9e-64477b56d533","title":"[Watchdog] Fix: [Atlas] Complete KG edge extraction from remaining (10 abandons)","completed":"2026-04-29T05:54:32.680942+00:00","priority":98},{"id":"8ebfce57-bd88-42cb-ad1d-272352db0505","title":"[Atlas] Complete KG edge extraction from remaining ~760 debate sessions — multiply KG density 7x","completed":"2026-04-29T05:54:22.093009+00:00","priority":96},{"id":"09a8e9af-a760-41f2-ba97-b8131faa8fdf","title":"snap-create-1","completed":"2026-04-29T05:50:23.051748+00:00","priority":50},{"id":"92a8939d-6bf8-48d6-8f29-42fedc29960d","title":"snap-stats-1","completed":"2026-04-29T05:49:35.513956+00:00","priority":50},{"id":"9ee35cb4-78a7-4ffa-961f-ef2cc1b64e9a","title":"snap-complete-1","completed":"2026-04-29T05:46:51.346761+00:00","priority":50},{"id":"26f4a7d1-099f-4597-af66-64c235f3831c","title":"[Watchdog] Fix: [Atlas] Complete KG edge extraction from remaining (10 abandons)","completed":"2026-04-29T05:32:54.009743+00:00","priority":98},{"id":"fe54dc47-f7fe-4e31-97fc-cece106b92a7","title":"[Watchdog] Fix: [Atlas] Complete KG edge extraction from remaining (10 abandons)","completed":"2026-04-29T05:29:02.745753+00:00","priority":98},{"id":"dee9dcbe-ffda-41e7-9f71-d7fec09e3143","title":"[Watchdog] Fix: [Atlas] Complete KG edge extraction from remaining (10 abandons)","completed":"2026-04-29T05:12:39.865871+00:00","priority":98},{"id":"ba6154f4-b514-47ce-94fc-7f1fc9769e4e","title":"[Agora] Systematic prediction evaluation pipeline — auto-evaluate 200+ predictions for validated hypotheses","completed":"2026-04-29T05:06:54.414650+00:00","priority":95},{"id":"0fa30abc-da44-4981-a200-83d91371411e","title":"[Forge] Experiment outcome capture pipeline — add execution results to experiments and propagate to hypothesis confidence","completed":"2026-04-29T05:06:49.392075+00:00","priority":93},{"id":"8cfc2288-829d-40e1-9851-fb987c36b901","title":"[Forge] Execute experiment: Circadian gene expression effects of SD vs ketamine","completed":"2026-04-29T05:05:57.558056+00:00","priority":70},{"id":"2b9b4791-88ec-408e-84f8-349336a4d228","title":"[Forge] Execute experiment: Single-cell RNA-seq analysis of mouse testis development and meiotic defects","completed":"2026-04-29T05:05:47.221764+00:00","priority":70},{"id":"1fc400b5-b4be-4337-81f0-4efb29a1c0f6","title":"[Senate] Triage: broken debate_sessions FK (15 orphans)","completed":"2026-04-29T05:03:58.321123+00:00","priority":90},{"id":"03094ddf-2ada-406a-8f5a-66f490d883df","title":"[Agora] Generate falsifiable predictions for 25 hypotheses with none","completed":"2026-04-29T05:02:50.681330+00:00","priority":92},{"id":"1c8911cb-da8b-480a-b626-374497dffb7d","title":"[Atlas] Validated hypothesis wiki reports — publish 106 validated findings as structured wiki pages with evidence chains","completed":"2026-04-29T05:00:48.540241+00:00","priority":94},{"id":"fc9309fd-03db-4d7a-b15c-9767db2d3fcb","title":"[Atlas] Neurodegeneration target dossier pipeline — build structured dossiers for 185 targets","completed":"2026-04-29T04:58:39.296871+00:00","priority":96},{"id":"5ef91ecb-3650-4a1a-a681-f27b70730383","title":"[Senate] Triage: [Watchdog] Fix: [Agora] Add PubMed evidence to 20 hypotheses lacki (11","completed":"2026-04-29T04:56:03.015834+00:00","priority":70},{"id":"7b90aa0b-2a29-4c41-8368-924fbe2d23e4","title":"[Senate] Triage: [Forge] Power-calc service: sample-size + effect-size estimator for an","completed":"2026-04-29T04:55:57.418475+00:00","priority":70}],"vm":{"mem_total_gb":61.4,"mem_used_gb":8.3,"mem_pct":13.5,"disk_total_gb":1982.8,"disk_used_gb":405.8,"disk_pct":20.5,"load_1m":7.02,"load_5m":5.72,"load_15m":4.96,"cpu_count":8,"uptime_hours":34.6},"agents":{"active_worktrees":31,"worktree_branches":["task-aa1c8ad8-f157-4e03-941d-9fdb76dbb12a","task-03094ddf-2ada-406a-8f5a-66f490d883df","task-fd5fe016-1581-4f74-8c67-0c7430987296","task-c69b704f-09e9-4baa-b46b-c407015f0f8f","task-11d9d78b-9f47-43e5-aa8d-fecb2ce1fb6f","task-16eb93d0-9277-4eef-893a-d659349b2def","task-e352460b-2d76-4d02-8057-b255dbcba532","task-f4f09ad5-2315-46fe-a526-fa6ab92dfe23","task-98c6c423-6d32-4090-ae73-6d6ddae3d155","task-d924270b-4ef8-4969-8877-9dd8b34d0239","task-5b88ec15-2824-432e-9d49-eb34e2c92cfe","task-1544b567-7581-467f-9a9e-1605913998b9","task-809da8ca-1ede-483e-8e5a-49553f4bfd1d","task-80ffb77b-8391-493c-8644-37086c8e2e3c","task-2310c378-ea0e-4bde-982e-cb08cc40be96","task-3ff23dae-528b-4aa3-aa23-495cf6ad1a16","task-38ad20f0-4d04-4af7-ad91-fdefac60b6e1","task-a67a9ff1-52c1-4dfc-bdf7-7b379af21706","task-2c4b95b0-39d3-4ae8-b067-f9ae1241aec3","task-2f7e1600-fc6c-46fe-b008-affd3aeee432"],"process_count":7,"autonomous_agents":0,"interactive_sessions":7,"orchestra_slots_alive":0,"orchestra_slots_total":13,"orchestra_concurrency":10},"db_stats":{"analyses":478,"hypotheses":2362,"kg_edges":710066,"wiki_pages":17790,"papers":29566,"skills":700,"debates":882,"analysis_queue_depth":0,"active_analyses":2},"watchdog":{"monitored_count":0,"grace_count":0,"kills_last_24h":0,"recent_events":[]},"backup_health":{"status":"critical","color":"#ef4444","latest_age_h":193.76,"backup_count":116},"quest_reviews":[{"quest_name":"Other","task_count":13,"total_tokens":650000.0,"total_cost":1.95,"output_score":11230.6089,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Atlas","task_count":9,"total_tokens":450000.0,"total_cost":1.35,"output_score":8150.2824,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"UI","task_count":9,"total_tokens":450000.0,"total_cost":1.35,"output_score":32033.6644,"efficiency_rating":"efficient","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Exchange","task_count":6,"total_tokens":300000.0,"total_cost":0.9,"output_score":22118.827,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Forge","task_count":5,"total_tokens":250000.0,"total_cost":0.75,"output_score":1105.4208,"efficiency_rating":"waste","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Senate","task_count":3,"total_tokens":150000.0,"total_cost":0.45,"output_score":22418.2133,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Demo","task_count":3,"total_tokens":150000.0,"total_cost":0.45,"output_score":20240.32,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Artifacts","task_count":3,"total_tokens":150000.0,"total_cost":0.45,"output_score":1376.2,"efficiency_rating":"waste","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Resources","task_count":2,"total_tokens":100000.0,"total_cost":0.3,"output_score":1657.27,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Arenas","task_count":1,"total_tokens":50000.0,"total_cost":0.15,"output_score":1447.86,"efficiency_rating":"normal","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Gaps","task_count":1,"total_tokens":50000.0,"total_cost":0.15,"output_score":1442.66,"efficiency_rating":"waste","reviewed_at":"2026-04-20T16:13:25.817325+00:00"},{"quest_name":"Search","task_count":1,"total_tokens":50000.0,"total_cost":0.15,"output_score":64298.9,"efficiency_rating":"efficient","reviewed_at":"2026-04-20T16:13:25.817325+00:00"}]}