Shared Selective Persistent Memory for Agentic LLM Systems
Apple paper: keep specs, schemas, tool config, and constraints; drop session traces. Selective shared memory beat both no memory and full history.
Knowledge catalogue
Apple paper: keep specs, schemas, tool config, and constraints; drop session traces. Selective shared memory beat both no memory and full history.
Social and low-evidence quarantine moved to raw as rejected. 168 high-specificity articles remain in inbox; four sources were filed.
Agent Seer synthesizes graded multi-turn tool-use eval scenarios from MCP specs alone—no examples, live tools, or domain tuning.
Apple derives a repetition-aware mixture scaling law showing scarce target data can be reused ~15–20× when mixed with generic corpora.
Apple ML shows LLM world-model simulation can generate API-agent training trajectories from specs alone—no executable environments required.
Vercel’s open-source agent framework — durable sessions, sandboxes, HITL, tools/skills as files; positioned as Next.js for agents.
SAG retrieves structured multi-hop evidence via event-entity hyperedges and query-time joins, without a maintained global knowledge graph.
AD2-Bench diagnoses multimodal reasoning via a Chain of Evidence, showing failures from spatial ambiguity and semantic uncertainty in adverse urban scenes.