# arXiv Scout Coverage — 2026-08-02_00-05

- Queries run: 14
- Total pages fetched: 18
- Total entries seen: 900
- Total unique papers kept before cross-query dedupe: 900
- Total duplicates skipped within queries: 0

## Per-query coverage

### primary-cs.AI
- Query: `cat:cs.AI`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 0
- Entries seen: 0
- Unique papers kept: 0
- Duplicates skipped: 0
- Newest seen: n/a
- Oldest seen: n/a
- Stop reason: fetch_failed
- Errors:
  - page=1 start=0 error=fetch_failed

### primary-cs.LG
- Query: `cat:cs.LG`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 0
- Entries seen: 0
- Unique papers kept: 0
- Duplicates skipped: 0
- Newest seen: n/a
- Oldest seen: n/a
- Stop reason: fetch_failed
- Errors:
  - page=1 start=0 error=fetch_failed

### primary-cs.CL
- Query: `cat:cs.CL`
- Backfill window: 72h
- Page size: 100
- Max pages: 5
- Pages fetched: 0
- Entries seen: 0
- Unique papers kept: 0
- Duplicates skipped: 0
- Newest seen: n/a
- Oldest seen: n/a
- Stop reason: fetch_failed
- Errors:
  - page=1 start=0 error=fetch_failed

### topic-world-model
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("world model" OR world-model)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:57:41+00:00
- Oldest seen: 2026-07-27T22:12:45+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:57:41+00:00 oldest=2026-07-29T10:47:52+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-29T10:14:02+00:00 oldest=2026-07-27T22:12:45+00:00 rate_limited=False status=200

### topic-agent
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND agent`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:59:11+00:00
- Oldest seen: 2026-07-29T17:50:39+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:59:11+00:00 oldest=2026-07-30T09:47:58+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-30T09:41:08+00:00 oldest=2026-07-29T17:50:39+00:00 rate_limited=False status=200

### topic-tool-use
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("tool use" OR tool-use)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:59:56+00:00
- Oldest seen: 2026-07-28T07:58:07+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:59:56+00:00 oldest=2026-07-28T07:58:07+00:00 rate_limited=False status=200

### topic-memory
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND memory`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:59:56+00:00
- Oldest seen: 2026-07-27T15:36:18+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:59:56+00:00 oldest=2026-07-29T14:47:19+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-29T14:15:29+00:00 oldest=2026-07-27T15:36:18+00:00 rate_limited=False status=200

### topic-reasoning
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND reasoning`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:57:41+00:00
- Oldest seen: 2026-07-29T04:37:59+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:57:41+00:00 oldest=2026-07-30T05:39:01+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-30T04:34:27+00:00 oldest=2026-07-29T04:37:59+00:00 rate_limited=False status=200

### topic-llm
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND (LLM OR "large language model" OR "large language models")`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:58:58+00:00
- Oldest seen: 2026-07-30T05:08:15+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:58:58+00:00 oldest=2026-07-30T11:50:10+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-30T11:49:46+00:00 oldest=2026-07-30T05:08:15+00:00 rate_limited=False status=200

### topic-quantization
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND quantization`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:43:36+00:00
- Oldest seen: 2026-07-21T08:58:36+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:43:36+00:00 oldest=2026-07-21T08:58:36+00:00 rate_limited=False status=200

### topic-open-source
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("open source" OR open-source)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-07-30T16:41:55+00:00
- Oldest seen: 2026-07-28T15:42:53+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T16:41:55+00:00 oldest=2026-07-28T15:42:53+00:00 rate_limited=False status=200

### topic-self-improvement
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("self-improvement" OR "self improvement")`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 1
- Entries seen: 50
- Unique papers kept: 50
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:34:01+00:00
- Oldest seen: 2026-06-29T14:20:47+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:34:01+00:00 oldest=2026-06-29T14:20:47+00:00 rate_limited=False status=200

### topic-fine-tuning
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND ("fine-tuning" OR finetuning OR finetune)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:57:10+00:00
- Oldest seen: 2026-07-27T15:18:06+00:00
- Stop reason: cutoff_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:57:10+00:00 oldest=2026-07-29T13:23:30+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-29T12:23:15+00:00 oldest=2026-07-27T15:18:06+00:00 rate_limited=False status=200

### topic-benchmark
- Query: `(cat:cs.AI OR cat:cs.LG OR cat:cs.CL) AND (benchmark OR benchmarks)`
- Backfill window: 96h
- Page size: 50
- Max pages: 2
- Pages fetched: 2
- Entries seen: 100
- Unique papers kept: 100
- Duplicates skipped: 0
- Newest seen: 2026-07-30T17:59:56+00:00
- Oldest seen: 2026-07-30T05:41:44+00:00
- Stop reason: max_pages_reached
- Pages:
  - page 1 start=0 entries=50 newest=2026-07-30T17:59:56+00:00 oldest=2026-07-30T12:17:31+00:00 rate_limited=False status=200
  - page 2 start=50 entries=50 newest=2026-07-30T12:06:53+00:00 oldest=2026-07-30T05:41:44+00:00 rate_limited=False status=200

## Priority summary
- High-priority papers: 478
- Priority threshold: 4
- Topic hits:
  - agent: 100
  - benchmark: 100
  - fine-tuning: 100
  - llm: 100
  - memory: 100
  - reasoning: 100
  - world-model: 100
  - open-source: 50
  - quantization: 50
  - self-improvement: 50
  - tool-use: 50
- Top papers:
  - 22.0 :: MANTA: Multi-Agent Network Topology Adaptation for Self-Evolving Multi-Agent Systems [agent, benchmark, llm, reasoning, self-improvement, tool-use]
  - 17.0 :: PAUSE: A User-Centric Benchmark for Personal AI Assistants in Unified Service Environments [agent, reasoning, tool-use, world-model]
  - 16.0 :: Can Agents Deceive? Evaluating Reasoning and Deception in ParliamentBench using a Social Deduction Game [agent, benchmark, llm, open-source, reasoning]
  - 16.0 :: MMHBench: A Multi-Perspective Benchmark for Mental Health Understanding in Long-Form Videos [agent, benchmark, llm, open-source, reasoning]
  - 16.0 :: ChronoMem: Version Control and Semantic Rollback for Large Language Model Agent Memory [agent, benchmark, llm, memory, open-source]
  - 16.0 :: Albilich: Steerable Proof-State Orchestration for LLM-Based Mathematical Research with CAS Integration [agent, benchmark, llm, open-source, reasoning]
  - 16.0 :: Rehearse: Stepping Back from the Confidence Cliff in Self-Improving Autoresearch [agent, llm, memory, self-improvement]
  - 16.0 :: See2Think: Do Multimodal Models Really Use Intermediate Visual States? [open-source, reasoning, tool-use, world-model]
  - 15.0 :: OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models [agent, benchmark, reasoning, world-model]
  - 15.0 :: MemHarness: Memory Is Reconstructed, Not Replayed [agent, llm, memory, reasoning]
  - 15.0 :: IndustryForge-27B: A Domain-Enhanced Multimodal Foundation Model for Industrial CAD [agent, benchmark, fine-tuning, self-improvement]
  - 15.0 :: ARES: Adaptive Reasoning-Effort Steering for PPA- and Cost-Aware RTL Optimization with LLM Agents [agent, llm, memory, reasoning]
  - 15.0 :: An Empirical Study of Coordination Mode as the First-Class Citizen in From-Scratch Multi-Agent Coding [agent, benchmark, reasoning, world-model]
  - 15.0 :: DeepResearch Agent System [agent, open-source, reasoning, tool-use]
  - 14.0 :: LEDGERMIND: Provenance-Constrained Multimodal Agentic Reasoning with a Structured Evidence Ledger [agent, benchmark, reasoning, tool-use]
  - 14.0 :: DataClawEval: A Benchmark for Data Engineering Agents in Real Industrial Harness [agent, benchmark, llm, world-model]
  - 13.0 :: ORCA-bench: How Ready Are Language Model Agents for Oncall? [agent, benchmark, llm, reasoning]
  - 13.0 :: ObjectStream: Latent Objects as Memory Anchors for Streaming Video Understanding [benchmark, llm, memory, reasoning]
  - 13.0 :: SKILL-KD: Contrastive Skill Distillation for LLM Agents [agent, benchmark, llm, memory]
  - 13.0 :: ClawTrack: Towards Trace-Level Evaluation and Improvement of Real-World Autonomous Agents [agent, benchmark, llm, reasoning]