[Senate] Holistic Task Prioritization & Self-Governance — Quest Spec

[Senate] Holistic Task Prioritization & Self-Governance — Quest Spec

Quest Type: Recurring Layer: Senate (cross-cutting: all layers) Priority: 90

Goal

Periodically review all open tasks across SciDEX and adjust priorities holistically based on system state, strategic alignment, resource constraints, and progress signals. This quest is the "central planning" function that ensures the agent swarm works on the most valuable things at any given time. The long-term vision is that SciDEX's own market and debate mechanisms (from the economics quest) are used to prioritize development tasks — the system governs its own evolution.

Scope: Quests AND Tasks, Holistically

This quest does not just re-rank individual tasks. It plans holistically across the entire portfolio of quests and tasks:

• Quest-level review: Are the right quests active? Are quest priorities balanced across layers? Should a quest be paused because its dependencies haven't delivered yet? Are two quests pulling in opposite directions?
• Cross-quest dependency mapping: The contributor network quest depends on the economics quest's threaded comments and generalized markets. The paper figures quest feeds into the economics quest's artifact markets. The prioritization quest itself depends on economics for market-driven Phase 2. These dependencies determine sequencing.
• Strategic rebalancing: If the system is healthy but the economics infrastructure is lagging, shift priority toward economics workstreams. If the site has broken pages, temporarily boost UI/health tasks above feature work. If a new external agent wants to onboard but the contributor network isn't ready, escalate that quest.
• Resource allocation across quests: With N agents available, which quests should get attention this cycle? A quest with 5 ready batch tasks and 0 blockers should get more agents than a quest whose next step is blocked.

Every recurring quest spec should reference this quest as the coordinator that may adjust its priority or sequencing based on the global picture.

Acceptance Criteria

Prioritization Framework

Phase 1: Rule-Based (immediate)

Score each open task on these dimensions:

Dimension	Weight	Signal Source
User Impact	25%	Does it fix a broken page? Improve UX?
Strategic Alignment	20%	Which layer does it advance? How central?
Dependency Unlocking	20%	How many other tasks are blocked by this?
Staleness Penalty	15%	How long since last progress? (decays priority)
Effort Estimate	10%	Quick wins get boosted (low effort, high value)
System Health	10%	Is the site currently broken in related areas?

def holistic_priority(task):
    impact = estimate_user_impact(task)        # 0-1
    alignment = layer_alignment_score(task)    # 0-1, based on current strategic focus
    unlocking = count_blocked_tasks(task) / 10  # normalized
    staleness = days_since_progress(task) / 30  # penalty, capped at 1
    effort_inv = 1.0 / max(estimated_hours(task), 1)  # quick wins boost
    health = related_page_errors(task)          # 0-1 if related pages are broken
    
    raw = (0.25 * impact + 0.20 * alignment + 0.20 * unlocking
           - 0.15 * staleness + 0.10 * effort_inv + 0.10 * health)
    return int(raw * 100)  # scale to 0-100 priority

Phase 2: Market-Informed (after economics quest delivers)

Once the generalized market framework exists, create a Task Market:

• Each open task gets a market price (0-1)
• Agents trade on tasks they believe are most valuable
• Price = community-assessed priority
• Token bounties can be placed on tasks
• Debate sessions can be triggered to argue for/against task importance

Task bounty pricing to agents: Tasks in the quest queue should carry a token bounty that reflects their market price, complexity, and strategic value. Agents choose which tasks to work on partly based on bounty — higher-value tasks attract more capable agents. This creates a natural labor market:

• Bounty = f(market_price, estimated_effort, urgency_bonus) — a task that the market prices highly AND is urgent AND is quick to complete carries the highest bounty-per-hour
• Agents bid on tasks: Instead of round-robin assignment, agents can express preference for tasks they're well-suited for, staking tokens to claim them (skin in the game — if they don't deliver, they lose the stake)
• Completion rewards scale with quality: Base bounty paid on completion, bonus multiplier for high-quality delivery (as assessed by reviewers or market movement)
• Specialization premium: Tasks requiring rare capabilities (e.g., molecular analysis, clinical trial expertise) carry higher bounties to attract specialized agents
• Open to all contributors: Bounty-priced tasks are visible to the entire contributor network — external agents, human participants, anyone registered. This is a key onboarding path: a new contributor browses open bounties, picks one they can solve, delivers, earns tokens and reputation. The same mechanism that prices knowledge gap challenges to the public also prices development tasks to the community. The task queue becomes a public marketplace where anyone can contribute

Task Market Integration:
1. New task created → auto-listed in task market at price 0.5
2. Agents review and trade (buy = "this is important", sell = "deprioritize")
3. Prioritization quest reads market prices as one input signal
4. High-priced tasks get priority boost, low-priced get deprioritized
5. Completed tasks → market resolves → traders rewarded/penalized

Phase 3: Debate-Driven (future)

Use Agora debates to argue about priorities:

• When priority is contested (big disagreement in task market), trigger a debate
• Theorist argues for the task, Skeptic argues against
• Expert assesses feasibility, Synthesizer produces a recommendation
• Debate quality score weights the final priority adjustment

Quest-Level Review (Each Run)

Before reviewing individual tasks, assess the full quest portfolio:

1. Quest Dependency Map

[Senate] Holistic Task Prioritization (P90) — THIS QUEST
    ├── depends on: [Exchange] Economics quest (for Phase 2 market-driven prioritization)
    └── coordinates: ALL other quests

[Exchange] Economics, Markets & Incentive Ecology (P88)
    ├── WS1: Threaded comments & voting (P85) ← foundational, unblocked
    ├── WS2: Generalized market framework (P85) ← foundational, unblocked
    ├── WS3: Agent reputation & tokens (P80) ← depends on WS1 (vote weights)
    ├── WS4: Market proposal governance (P80) ← depends on WS2 + WS3
    └── WS5: Comment-driven price signals (P75) ← depends on WS1 + WS2

[Senate] Contributor Network & External Agents (P86)
    ├── depends on: Economics WS1 (comments on profiles)
    ├── depends on: Economics WS2 (agent markets)
    ├── depends on: Economics WS3 (reputation system)
    └── Phase 1 (role activation, profiles) is unblocked

[Atlas] Paper Figures Extraction (P82)
    ├── Phase 1 (PMC API) is unblocked
    ├── Phase 4 depends on: Economics WS2 (figure markets)
    └── feeds into: Agora debates (multimodal evidence)

[Forge] Expand tool library (P50) — ongoing
[UI/Senate/Agora] CI quests — ongoing health maintenance

2. Quest-Level Questions to Answer Each Run

• Which quests have unblocked work that no agent is doing? → Boost priority
• Which quests are blocked on another quest's deliverable? → Focus on the blocker
• Are quest priorities balanced across layers? (If all P90 quests are Senate, Atlas is starved)
• Is any quest scope-creeping beyond its spec? → Split or refocus
• Should any quest be paused (temporarily deprioritized to 0) to focus resources?
• Are two quests conflicting (building competing implementations of the same thing)?
• Has the strategic context changed (new user priority, external event, system health issue)?

3. Rebalancing Actions

Situation	Action
Quest A blocks Quest B, Quest B is higher priority	Boost Quest A above Quest B
Quest has been active 2+ weeks with no completed batch tasks	Review scope, consider splitting
Two quests overlap significantly	Merge into one, close the other
Layer is underrepresented in top-20 priorities	Boost that layer's quests
System health is degraded	Temporarily boost all CI/health quests, pause feature work
New user-requested quest arrives	Assess where it fits in dependency graph, set priority accordingly

Operational Procedure (Each Run)

# 1. Get system state
scidex status
orchestra task list --project SciDEX --status open --format json

# 2. Check site health (broken pages boost related task priority)
python3 link_checker.py --quiet
for page in / /exchange /senate /forge /graph /analyses/ /gaps; do
    curl -so/dev/null -w'%{http_code}' "localhost$page"
done

# 3. Identify stale tasks (no spec work log update in 7+ days)
# Scan docs/planning/specs/ for tasks with old last work log entry

# 4. Identify duplicates
# Group tasks by similar titles/descriptions, flag overlaps

# 5. Compute holistic priorities
# Apply scoring formula, compare with current priorities

# 6. Adjust
orchestra task update --id $TASK_ID --priority $NEW_PRIORITY
# Close stale/duplicate tasks with explanation

# 7. Report
# Append prioritization summary to this spec's work log

Stale Task Handling

Days Since Progress	Action
7-14	Flag as "needs attention" — boost priority by 5 to attract agents
14-30	Deprioritize by 10 — if no one picked it up, it may not be important
30+	Propose closing — add comment explaining why, close if no objection in 48h
Blocked	Identify blocker, escalate the blocking task's priority

Duplicate Detection

Tasks are potential duplicates if:

• Title similarity > 80% (Levenshtein or token overlap)
• Same layer prefix AND similar keywords
• Both reference the same spec file or code file

Action: Merge the lower-priority duplicate into the higher one, append its description.

Integration with Economics Quest

This quest and the economics quest form a virtuous cycle:

Economics quest builds market/voting/token infrastructure

This quest uses that infrastructure to prioritize all tasks (including economics tasks)

Better prioritization → faster economics delivery → better prioritization tools

The task market becomes the most meta of all SciDEX markets: the system pricing its own development priorities.

Approach

Create this quest as recurring in Orchestra

Implement priority scoring function (start with rule-based Phase 1)

Run first prioritization review — adjust top 20 misaligned tasks

Set up periodic execution (CI-style, like other recurring quests)

As economics quest delivers markets: integrate task market signals (Phase 2)

As Agora evolves: add debate-driven prioritization (Phase 3)

Work Log

2026-04-04 05:24 PDT — Slot 11 — Prioritization Review

System Health: ✅ Good — all pages 200, API healthy Portfolio: 18 running, 7 open — well distributed Key Findings:

• KG edge coverage: 24% (14/58 completed analyses have edges) → Atlas CI at P92 is priority
• Quest 17 (Artifacts) had no active CI task → created [Artifacts] CI task (73ee83a5) at P85
• 19 failed analyses: transient LLM parse errors, no systemic bugs
• Demo (Quest 16) is well-covered with 5 active tasks

Actions: Created Artifacts CI recurring task. No priority adjustments needed.

2026-04-04 10:38 PT — Slot 2 — Prioritization Review

System Health: ✅ Good

• API: Active and responding (200s on all endpoints)
• Services: agent, bridge, nginx, linkcheck, neo4j all green
• Issues: improve and forge-improve inactive (expected for non-essential services)
• All key pages load: /, /exchange, /senate, /forge, /graph, /analyses/, /gaps, /atlas.html, /how.html

Task Portfolio Analysis:

• Total open tasks: 2
• Total running tasks: 32
• Total in_progress: 1
• Total blocked: 0
• Tasks with errors: 4 (all "Worker lease expired; requeued" - expected behavior)

Stale Task Check: ✅ None

• All running/open tasks updated within last 2 days
• No tasks older than 7 days without progress

Quest-Level Balance Review:

Quest	Priority	Task Count	Status
Demo	P99	503	✅ Active - 4 running tasks
Search	P95	195	✅ 1 running task
Artifacts	P90	414	✅ Active - 5 running tasks
Agora	P90	77	✅ 1 running task
Exchange	P85	851	✅ 2 running tasks
Forge	P80	408	✅ 2 running tasks
Atlas	P80	463	✅ 3 running tasks
UI	P75	923	✅ 4 running tasks
Senate	P60	455	✅ 1 running task

Priority Adjustments: None needed

• Demo (P99) and Artifacts (P90) are getting appropriate attention per QUESTS.md
• Running tasks are well-aligned with strategic priorities
• No blocked tasks to escalate
• No stale tasks to close
• No duplicate quests detected

Recommendations:

Senate at P60 is intentional per QUESTS.md - governance layer runs in background

Economics at P70 is appropriate - Phase 2 market-driven prioritization not yet needed

Current task distribution is healthy - no rebalancing required

2026-04-04 04:50 PDT — Slot 9 — System Health Recovery + Task Review

System Health: ⚠️ → ✅ Recovered

• Fixed critical API startup bug: missing from typing import Any, List import
• API now stable and serving requests (200s on /vision, /exchange, /analyses/)
• Database: 76 analyses, 180 hypotheses, 180 KG edges, 1 open gap

Task Portfolio:

• Open: 24 tasks | Running: 20 tasks | Blocked: 0 tasks
• Priority distribution:

- P90+: 6 tasks (high-value work active)
- P70-89: 13 tasks (healthy mid-tier)
- P42: 1 task (low-priority maintenance)

Issues Identified:

8 duplicate/transient error tasks at P70 - All auto-generated from link checker during API downtime

- Titles: "Server Failure", "Unavailability", "500 errors", "Dynamic Content Not Generated"
- Root cause: API was crashlooping due to missing imports
- IDs: 667ede08, 51bbc267, 0514eb5b, 918c2de0, 673689a9, a9bbf4cf, c172e5fc, a5fdc03f
- Recommendation: Monitor for auto-resolution; if persistent after 24h, manually close as duplicates

Quest-Level Balance: ✅ Well-aligned

• Demo (P99): Active work ongoing
• Artifacts (P90): Active work ongoing
• All layers have appropriate coverage
• No dependency blockers identified

Actions Taken:

• Fixed API bug → system health restored
• No priority adjustments needed (distribution is healthy)

Stale Task Check: ✅ None found

• All running/open tasks updated within last 48 hours

2026-04-06 UTC — Prioritization Review [task:b4c60959-0fe9-4cba-8893-c88013e85104]

System Health: ✅ Good

• API: Active, all pages 200 (/ redirects 302 → expected)
• DB: 118 analyses, 314 hypotheses, 688,411 KG edges, 68/70 open gaps
• Services: healthy

Task Portfolio: 23 open → 29 open after this review | 3 running

• Distribution: P88-P99 (10 tasks), P80-87 (9 tasks), P72-79 (4 tasks)
• Layer coverage: Agora×2, Atlas×3, Demo×2, Economics×1, Exchange×3, Forge×2, Gaps×2, Senate×3, System×1, UI×3, Wiki×1

Critical Finding: 14 high-priority quests (P88-P95) with NO open tasks

Quest	Priority	Gap
Epistemic Rigor	P95	No tasks — 8-task architecture defined, never started
Search	P95	1 open task in spec, not in Orchestra
Experiment Extraction	P93	No tasks in queue
Autonomous Engines	P93	No tasks in queue
Evolutionary Arenas	P92	Recent work (Apr 6), no open tasks needed now
Artifact Quality Markets	P92	Full spec with 7 tasks, none in Orchestra
Real Data Pipeline	P92	Spec tasks use P3-P5 scale, likely intentionally lower priority
Capital Markets	P91	Full spec with 6 tasks, none in Orchestra
Artifact Debates	P91	Full spec with 6 tasks, none in Orchestra
Analysis Sandboxing	P90	Spec tasks use P3-P5 scale
Tool Playground	P88	No tasks
Schema Governance	P88	Full spec with 6 tasks, none in Orchestra
Open Debates	P88	Spec tasks use P2-P3 scale
Resource Governance	P88	Spec tasks use P4-P5 scale

Quest Priority Alignment (existing open tasks): ✅ Well-aligned

• All 23 existing open tasks within ±8 points of their quest priority
• No major misalignments identified

Actions Taken — 6 tasks created for uncovered quests:

dca29466 [Agora] epi-01-PRED P92 — hypothesis_predictions table (Epistemic Rigor quest P95)

53647a76 [Exchange] exch-qm-05-GATE P91 — Quality gates for all artifact types (Artifact Quality Markets P92)

70907ae9 [Agora] agr-ad-01-TARG P91 — Generalize debate targeting (Artifact Debates P91)

360e2578 [Exchange] exch-cm-01-LEDG P91 — Token ledger double-entry accounting (Capital Markets P91)

f557e9ea [Search] P90 — Ensure /search end-to-end with all content types (Search P95)

47b17cbf [Senate] sen-sg-01-SREG P88 — Schema registry with versioning (Schema Governance P88)

Priority Adjustments: None to existing tasks Stale Tasks: None (all tasks updated within 72h) Duplicate Tasks: None detected

2026-04-06 UTC (run 2) — Quest Assignment Fixes + New CI Tasks [task:b4c60959-0fe9-4cba-8893-c88013e85104]

System Health: ✅ Good

• API: Active (302 redirect → expected)
• DB: 119 analyses, 314 hypotheses, 688,411 KG edges, 15,732 papers
• Total open/running tasks: 35

Issues Found — Quest Assignment Mismatches:
The 6 tasks created in the previous run were assigned to parent-layer quests instead of their specific sub-quests. Fixed:

• dca29466 epi-01-PRED → q-epistemic-rigor (was: Agora)
• 53647a76 exch-qm-05-GATE → q-artifact-quality-markets (was: Exchange)
• 70907ae9 agr-ad-01-TARG → q-artifact-debates (was: Agora)
• 360e2578 exch-cm-01-LEDG → q-capital-markets (was: Exchange)
• 47b17cbf sen-sg-01-SREG → q-schema-governance (was: Senate)

New Tasks Created:

ID	Quest	Priority	Title
4132c477	q-tool-playground (P88)	P87	forge-tp-02-EXPAND: Add 6+ tools to playground (ClinVar, HPA, GWAS, Open Targets)
607558a9	q-evolutionary-arenas (P92)	P91	CI: Run daily King of the Hill tournament
8f7dc2dc	q-experiment-extraction (P93)	P92	CI: Verify extraction quality metrics + extract from new papers

P88+ Quest Coverage After This Run:

• P95 Epistemic Rigor: 1 open ✓
• P95 Search: 1 open ✓
• P93 Experiment Extraction: 1 open (new CI) ✓
• P93 Autonomous Engines: 0 open — debate engine is recurring every-30-min (completed = scheduled) ✓
• P92 Artifact Quality Markets: 1 open ✓
• P92 Evolutionary Arenas: 1 open (new CI) ✓
• P92 Real Data Pipeline: 0 open — spec uses P3-P5 scale intentionally ✓
• P91 Artifact Debates: 1 open ✓
• P91 Capital Markets: 1 open ✓
• P90 Analysis Sandboxing: 0 open — spec uses P3-P5 scale intentionally ✓
• P88 Code Health: 1 open ✓
• P88 Schema Governance: 1 open ✓
• P88 Tool Playground: 1 open (new) ✓
• P88 Open Debates: 0 open — spec uses P2-P3 scale intentionally ✓
• P88 Resource Governance: 0 open — spec uses P4-P5 scale intentionally ✓

Stale Tasks: None detected Duplicate Tasks: None detected Priority Adjustments: None needed (distribution remains healthy)

2026-04-06 UTC (run 3) — Validation + Final State Confirmation [task:b4c60959-0fe9-4cba-8893-c88013e85104]

System Health: ✅ All pages 200 (port 8000)

• DB: 119 analyses, 314 hypotheses, 688,411 KG edges, 15,732 papers
• Active tasks: 34 (30 open + 4 running)

Key Finding: Epistemic Rigor Quest (P95) is FULLY COMPLETE

• All 15 tasks done — dca29466 (epi-01-PRED) created in run 1 was a duplicate of pre-existing tasks and was cleaned up by orphan management. This is correct behavior.

Full Quest Coverage Audit (P88+):

Quest	P	Status	Open	Notes
Epistemic Rigor	95	active	0	15/15 done — COMPLETE ✓
Search	95	active	1	f557e9ea ✓
Experiment Extraction	93	active	0	8f7dc2dc running ✓
Autonomous Engines	93	active	0	Covered by recurring CI tasks in other quests ✓
Evolutionary Arenas	92	active	1	607558a9 ✓
Artifact Quality Markets	92	active	1	53647a76 ✓
Capital Markets	91	active	1	360e2578 ✓
Artifact Debates	91	active	1	70907ae9 ✓
Agora	90	active	2	bf55dff6, e4cb29bc ✓
Artifacts	90	active	1	73ee83a5 ✓
Tool Playground	88	active	1	4132c477 ✓
Schema Governance	88	active	1	47b17cbf ✓
Code Health	88	active	1	2e7f22a0 ✓

Lower Priority Quests (P80-87):

• Exchange (P85): 3 active (running + 2 open) ✓
• Gap Factory (P85): 2 open ✓
• Agent Ecosystem (P82): 2 unaccounted tasks (cancelled, acceptable) — quest at P82 lower priority
• Forge (P80): 2 open ✓
• Atlas (P80): 3 open ✓
• Work Governance (P78): 1 unaccounted (cancelled, acceptable) ✓

Actions Taken: None — portfolio is healthy, no gaps identified. Stale Tasks: None detected Duplicate Tasks: None detected Priority Adjustments: None needed

2026-04-06 UTC (run 4) — Portfolio Health Audit [task:b4c60959-0fe9-4cba-8893-c88013e85104]

System Health: ✅ All pages 200 (/, /exchange, /senate, /forge, /graph, /gaps)

• API: Active and responding
• Services: healthy

Task Portfolio: 34 active (32 open + 2 running)

• P99: Demo×2, P98: Senate-CI×1, P92: ExperimentExtraction (running)×1
• P91: ArtifactQualityMarkets×1, ArtifactDebates×1, CapitalMarkets×1, Arenas×1
• P90: Agora×2, Artifacts×1, Search×1, Senate(this task, running)×1
• P88: CodeHealth×1, SchemaGovernance×1
• P87: ToolPlayground×1, P85-P72: 9 additional tasks

Quest Coverage Audit (P85+):

Quest	P	Active	Notes
Epistemic Rigor	95	0	COMPLETE — 15/15 done ✓
Search	95	1	f557e9ea ✓
Experiment Extraction	93	1	Running ✓
Autonomous Engines	93	0	Covered by recurring CI ✓
Evolutionary Arenas	92	1	607558a9 ✓
Artifact Quality Markets	92	1	53647a76 ✓
Real Data Pipeline	92	0	Spec intentionally P3-P5 scale ✓
Capital Markets	91	1	360e2578 ✓
Artifact Debates	91	1	70907ae9 ✓
Analysis Sandboxing	90	0	Spec intentionally P3-P5 scale ✓
Code Health	88	1	✓
Schema Governance	88	1	✓
Tool Playground	88	1	✓
Open Debates	88	0	Spec intentionally P2-P3 scale ✓
Resource Governance	88	0	Spec intentionally P4-P5 scale ✓
Exchange	85	3	✓
Gap Factory	85	2	✓
Market Participants	85	0	Justified — depends on Capital Markets (P91, open) completing first; spec tasks are P78-P82 ✓

P80-84 Quests:

• Agent Ecosystem (P82): 0 active — acceptable (was noted in run 3)
• Atlas (P80): 3 active ✓
• Forge (P80): 2 active ✓
• Crypto Wallets (P80): 0 active — future quest, depends on Capital Markets ✓

Task Errors: 10 tasks show "Worker lease expired; requeued" — normal recurring behavior, not concerning

Stale Tasks: None — all tasks updated 2026-04-06 Duplicate Tasks: None detected Priority Adjustments: None needed New Tasks Created: None — portfolio is complete and healthy

2026-04-12 UTC — Portfolio Scale-Up Review [task:dff08e77-5e49-4da5-a9ef-fbfe7de3dc17]

System Health: ✅ API active

• DB: 259 analyses (+46/24h), 347 hypotheses (+12/24h), 700,924 knowledge_edges (+12,513 since Apr 6), 3,259 gaps, 17,539 wiki pages, 134 debate sessions (13/24h), 6 datasets
• Economics: 6,426 reward events, 64 active token accounts, 54,187 SCI tokens

Task Portfolio: 106 active (85 open + 21 running) — up from 34 on Apr 6

Running Tasks (21):

• P97: Economics evolution (1f62e277, recurring), Data-driven hypotheses (ba0513b9, one_shot)
• P96: Agent credit pipeline (1a3464d6)
• P95: Reward emission (44651656), Funding allocator (5531507e)
• P94: Debate trigger CI (bf55dff6)
• P92: This task (dff08e77), Calibration slashing (c3a426dc)
• P91: Gap quality improvement (e40b93aa, one_shot)
• P90: Tool library expansion (f13a8747), Weekly retrospective (610c708a)
• P85: Debate engine slot 2 (e4ed2939)
• P60: Add /datasets page (24a2737e, one_shot)
• P50: Dedup scan no-prefix (b1205b5b — possible orphan duplicate of 7ffcac76)

Key Observations:

KG healthy & growing: knowledge_edges = 700,924 (+12,513 since Apr 6). Note: separate kg_edges table has 30 entries — newer schema artifact, not a regression.

Hypothesis debate gap: 202/347 (58%) hypotheses lack debates. Debate trigger CI (p94) is active — coverage will improve.

Gap overproduction: 3,202 new gaps in last 7 days. Gap identification tasks appropriately low at p30. Squad autoseed (p95) gates new squads on high-priority gaps — correct safety valve.

Economics pipeline healthy: All economics drivers running (p95-p97). Reward events, token accounts, calibration slashing all active.

Datasets nascent: 6 datasets — Atlas Dolt work (p92) continuing.

Possible zombie: b1205b5b "Dedup scan every 6h" runs at p50 with no layer prefix. Appears to duplicate 7ffcac76 "[Forge] Dedup scan every 6h" (open, p60). Monitor but do not cancel without investigation.

Priority Adjustments Made:

• 5ac80446 [Exchange] Agent reputation system: p87 → p90 — key economic primitive enabling differentiated rewards based on track record; raising priority ensures earlier execution

Stale Tasks: None — all tasks updated within expected windows Duplicate Tasks: b1205b5b flagged for monitoring (possible orphan of 7ffcac76) New Tasks Created: None — portfolio well-populated across all layers

2026-04-03 — Quest Created

• Designed 3-phase prioritization framework (rule-based → market → debate)
• Defined staleness handling and duplicate detection
• Identified integration path with economics quest
• Removed 100-task cap from AGENTS.md — organic growth with quality prioritization is better than artificial limits

File: task_prioritization_quest_spec.md

Modified: 2026-04-25 22:00

Size: 26.8 KB