Newshot
AI Still developing 中文 Event #01M7MC484WX9KX47BCDTM3M0A8

Show HN: AI Agent Infrastructure and Memory Compaction Specifications

Event overview(model summary, not the source's own text)

A Show HN submission titled "AI Agent Infrastructure and Memory Compaction Specifications" appeared alongside two new arXiv preprints on agent skills and agent memory, both announced as v1. The first preprint, SkillBuffBench (arXiv:2611.22017v1), introduces a benchmark for measuring what Agent Skills contribute across harnesses, models, skill access modes and memory, describing Agent Skills as file-based packages of procedural knowledge that a coding agent can discover and load on demand. The second, EvoDisco (arXiv:2611.22102v1), focuses on strategy discovery for evolving agent memory, arguing that how an agent's memory evolves during interaction matters as much as what is stored. The supplied material does not describe the contents of the Show HN post itself beyond its title, and neither

This summary was distilled by a model from the 3 reports below; it is not the source text. Treat the reports themselves as the facts.

How this story unfolded

Left: what it concluded — the key points the model distilled from the reports below, each one listing the reports it rests on. Right: what happened — ordered by time, each with a short summary so you can follow the thread without clicking through. The numbered links work both ways: tap a source under a point to jump to that report, tap the cited-by chip on a report to jump back to the point.

Key points6 items Model summary · ordered by when the evidence was published

  1. Point 01 11/17, 16:16

    A Show HN post titled "AI Agent Infrastructure and Memory Compaction Specifications" was submitted; the material gives no further detail on its contents.

  2. Point 02 11/18, 13:00

    SkillBuffBench is released as arXiv:2611.22017v1 under the new-announcement category.

  3. Point 03 11/18, 13:00

    SkillBuffBench measures what Agent Skills contribute across harnesses, models, skill access modes and memory.

  4. Point 04 11/18, 13:00

    Agent Skills are described as file-based packages of procedural knowledge that a coding agent can discover and load on demand.

  5. Point 05 11/18, 13:00

    EvoDisco is released as arXiv:2611.22102v1 and addresses strategy discovery for evolving agent memory.

  6. Point 06 11/18, 13:00

    EvoDisco's abstract holds that how an LLM agent's memory evolves during interaction is as important as what that memory contains.

ReportsBy publish time · newer further down

  1. November 17, 2026 · Tue
  2. November 18, 2026 · Wed
  3. +21h

    EvoDisco: Strategy Discovery for Evolving Agent Memory

    Original summary arXiv:2611.22102v1 Announce Type: new Abstract: Large language model agents accumulate experience in memory, and how that memory evolves during interaction is as important as what

    arXiv cs.AI Original ↗
    Cited by Point 05 Point 06 Cited
  4. +21h

    SkillBuffBench: Measuring What Agent Skills Contribute across Harnesses, Models, Skill Access Modes and Memory

    Original summary arXiv:2611.22017v1 Announce Type: new Abstract: Agent Skills, file-based packages of procedural knowledge that a coding agent can discover and load on demand, are now shipped by ev

    arXiv cs.AI Original ↗
  5. Still developing · later reports will keep extending this axis

Related reading

Distilling Safety Filters for Filter-free Multi-agent RL
arXiv:2611.19190v1 Announce Type: new Abstract: Reinforcement learning (RL) can teach robot teams to navigate shared spa
arXiv cs.RO 11/18/2026 Full text here
82pts
Evolving Infra: Agent-Adaptive LLM Serving at Scale
HN AI 11/17/2026 Full text here
70pts
Nectar launches an AI agent to automate creator discovery and outreach for brands
Nectar's new AI agent automates influencer marketing, helping brands discover creators, manage outreach and partnerships
TechCrunch AI 11/13/2026 Full text here
77pts
Anthropic says the key to agent memory is knowing what to forget, not how much to store
Anthropic studied how Claude retains knowledge across sessions. Selective notes beat exhaustive archives by a wide margi
The Decoder 11/16/2026 Full text here
74pts
Do Verifier Judgments Survive Abstraction? Measuring Verification Transport for LLM Agent Evaluation
arXiv:2611.18471v1 Announce Type: new Abstract: LLM-based verifiers are increasingly used to evaluate agent executions,
arXiv cs.CL 11/16/2026 Full text here
62pts