frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Memory system for AI agents with associations, forgetting, synthesis

https://github.com/jarimustonen/formative-memory
2•jari_mustonen•1h ago
I built an open-source memory plugin for OpenClaw agents, a biologically-inspired alternative to OpenClaw's built-in memory-core, with a take on what consolidation should do.

My background is in psychology, specifically personality psychology and how personality builds on top of memory. If we let an agent gather memories, it allows the agent to show features that look like a "personality". The basic architecture is to let the memory system form more complex memories from the initial facts that get remembered. To build this consolidation, memories are forgotten, and memories that are retrieved often get stronger. There's also a structure that builds associations, used both in retrieval and in spotting memories that are candidates for consolidation.

This is what consolidation here looks like:

- *Weighted bidirectional associations* between memory pairs, built from co-retrieval. Recall expands one hop along strong edges, so related memories surface even when they don't directly match the query. - *Reconsolidation ("coloring")* — when a newer memory contradicts an older one, the older memory is rewritten in light of the new context. - *Per-recall Hebbian strengthening* based on which injected memory the agent actually leaned on in the reply. - *Content-addressed identity* (SHA-256), so the same fact from two sources collapses to one object at write time.

Today it ships as an OpenClaw plugin (v0.5.5, MIT). The core is agent-agnostic by design and can be extended to other agents.

Repo: https://github.com/jarimustonen/formative-memory Docs: docs/architecture.md and docs/how-memory-works.md in-repo npm: formative-memory

Honest scope: first time I'm posting this publicly. We've been running it across four bots for a few weeks. I'd love feedback.

A representation model as a workaround for catastrophic forgetting in LLMs

https://gist.github.com/alexmorleyfinch/bc0ee96fb72e9aafbad045939b982efd
1•dreamforever•1m ago•0 comments

LCM: Lossless Context Management

https://arxiv.org/abs/2605.04050
2•Brajeshwar•3m ago•0 comments

The River Otter's Remarkable Comeback

https://www.rewildingmag.com/the-river-otters-remarkable-comeback/
1•surprisetalk•4m ago•0 comments

Show HN: After 15 years of lifting and spreadsheets, I built my own workout app

https://www.setsweat.com
1•josephmarkus•4m ago•0 comments

Reasons not to use Skype (2012)

https://stallman.org/skype.html
1•downbad_•5m ago•1 comments

Farmers get short-changed in our current food system

https://www.foodpolitics.com/2026/05/28473/
2•speckx•6m ago•0 comments

First Church of the Singularity: An AI Art Experiment

https://firstchurchofthesingularity.com/
2•FocusedPotato•7m ago•0 comments

Show HN: I built an SQL layer for a NoSQL database

2•cammasmith•7m ago•0 comments

Show HN: Linlore – Defending human existence, one question at a time

https://linlore.com/
2•calinf•8m ago•0 comments

Just a Man – Dave Lofts

https://www.youtube.com/watch?v=toqNuEE1zkw
2•keepamovin•9m ago•0 comments

Devenv 2.1: Nix with native support for zsh, fish, and nushell via libghostty

https://devenv.sh/blog/2026/05/07/devenv-21-nix-with-zsh-fish-and-nushell-via-libghostty/
2•domenkozar•10m ago•0 comments

Show HN: Open-source AI code review (AGPLv3, BYOK)

https://github.com/kodustech/kodus-ai
2•gamalinosqui•10m ago•0 comments

Show HN: AERF – receipts for AI agents, crypto-signed like cosign for evidence

https://github.com/aerf-spec/aerf
2•keertahacker•11m ago•0 comments

Gode Cookery – Authentic Medieval Recipes

http://www.godecookery.com/godeboke/godeboke.htm
2•Mr_Minderbinder•13m ago•0 comments

Show HN: Avoiding "if" makes Quicksort faster

https://easylang.online/blog/qsort
2•chrka•13m ago•0 comments

Show HN: Pomota – Pomodoro timer that forces breaks (Tauri, Rust+React)

https://github.com/p32929/pomota
2•heliskyr2•14m ago•0 comments

Comparing Sandboxing Approaches for AI Agents

https://www.docker.com/blog/comparing-sandboxing-approaches-ai-agents/
2•chmaynard•14m ago•0 comments

OpenAI's Data Agent and the S3 Gap

https://datachain.ai/blog/openai-data-agent-s3-gap
2•shcheklein•15m ago•0 comments

Git for AI Agents

https://github.com/regent-vcs/re_gent
4•doshay•15m ago•1 comments

How to design a closed-loop data platform from first principles

https://beyondthetraverse.substack.com/p/how-to-design-a-closed-loop-data
2•ericpsimon•16m ago•1 comments

What's Gone Wrong at GitHub?

https://leaddev.com/software-quality/whats-gone-wrong-at-github
3•chhum•16m ago•0 comments

NYMes – a modern alternative to classic NFTs

https://oc2mx.net/NYMe.html
2•Ch1ffr3punk•16m ago•1 comments

Encoding different messages at different video resolutions [video]

https://www.youtube.com/watch?v=UCBAqaT4SQc
2•exploraz•18m ago•0 comments

Taiwan Became the Most Perilous Geopolitical Chokepoint [video]

https://www.youtube.com/watch?v=lxmt7rz4DLQ
2•mooreds•18m ago•0 comments

Debt Has Always Been the Ruin of Great Powers. Is the U.S. Next?

https://www.wsj.com/politics/policy/debt-has-always-been-the-ruinof-great-powers-is-the-u-s-next-...
4•mooreds•18m ago•0 comments

The New El Niño Might Make 2027 the Hottest Year on Record

https://www.nytimes.com/2026/05/06/opinion/el-nino-climate.html
3•mooreds•19m ago•0 comments

The Buses Should Be Free

https://nicholasdecker.substack.com/p/the-buses-really-should-be-free
2•surprisetalk•19m ago•0 comments

Scientists Gave 'Aggressive' Fish Psychedelic Drugs. A Breakthrough Came Next

https://www.404media.co/fish-psilocybin-magic-mushrooms-study-psychedelics/
2•Brajeshwar•20m ago•0 comments

Cognitive Surrender – Addy Osmani

https://twitter.com/addyosmani/status/2052124873208799378
2•adrianthedev•20m ago•0 comments

Xfra AI

https://www.xfra.ai/
3•snickmy•21m ago•0 comments