First screen, captured 5 days ago. Open the video.

Agent memory needs its OpenTelemetry moment

What HN thinks

A thread about standardizing agent memory that mostly remembered to complain the article was written by an agent.

Too early to standardize
OpenTelemetry came after decades of experience; agent memory is still changing fast and premature standards would add friction and look like SOAP or SGML.
Memory hurts, use plain docs
Many find agent memory records one-off details and biases the model, preferring curated markdown developer docs in the repo or explicit instructions to write notes to a file.
The article reads as LLM slop
A large subthread criticized the post as AI-written; the author admitted Claude drafted it from his outline, and others pushed back on complaints that ignore the content.

Consensus: Most agree a formal memory standard is unnecessary or premature, and that simple markdown files in a directory or repo work well enough for now.

Recommended in the comments

OpenViking (tool): A commenter has run it self-hosted for over six months with their own LLM and embedding model.
Pangram (service): Recommended as a reliable sanity check for spotting AI-written web content.
recall ↗ (library): Its author says it just works in several production workflows.

Summarised by Claude from the comment thread 5 days ago.

Points: 171
Comments: 103
Best rank: #1
Posted: 2026-11-08 by mthines
Category: AI & ML

agent-memory opentelemetry mcp ai-coding-agents standards lorekit agents-md

All captures:

Full page

Captured 5 days ago. Click for the whole screenshot.

Looks similar

Screenshot of stork.ai: Apple capped engineers at a $200 daily AI budget

Apple capped engineers at a $200 daily AI budget (stork.ai)

60 points by GloVin 4 days ago | 76 comments

A Stork AI daily newsletter issue led by a report that Apple cut some engineers' daily AI spending cap to $200, plus other AI news items.

apple ai-costs llm-inference tokenization

Screenshot of lesswrong.com: I struggle to believe that AI R&D is ingenious

I struggle to believe that AI R&D is ingenious (lesswrong.com)

136 points by porridgeraisin 7 days ago | 155 comments

A LessWrong shortform comment by Arjun Panickssery arguing AI agents show poor research judgment in autonomous R&D side projects like a Scrabble engine and philosophy research.

ai-research-agents autonomous-r-and-d llm-capabilities lesswrong

Screenshot of huggingface.co: Kimi K3.1 Pro model weights are out

Kimi K3.1 Pro model weights are out (huggingface.co)

607 points by abrarthought 9 days ago | 174 comments

Hugging Face model card for Moonshot AI's Kimi K3.1 Pro, an open-weight 2.8T-parameter multimodal MoE model with 1M-token context, including architecture and benchmark results.

kimi moonshot-ai open-weights llm

Unlocking any car from X brand in most of the world (shubs.io)

578 points by infosecau 5 days ago | 317 comments

Screenshot of ryanprior.com: Nyquist Limit:The frequency of changes in tech is invalidating our understanding

Nyquist Limit:The frequency of changes in tech is invalidating our understanding (ryanprior.com)

50 points by evakhoury 9 days ago | 21 comments

Blog post by Ryan Prior arguing that LLM-driven rapid change in software exceeds the human 'sampling rate' of understanding, using the Nyquist limit as an analogy.

nyquist-limit sampling-theorem llm software-engineering