frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: MSAM – Memory system for AI agents that knows when it doesn't know

https://github.com/jadenschwab/msam
1•jadenschwab•1h ago
I was originally deploying OpenClaw and found their method of storing data in the workspace very inefficient and even with attempting a a hierarchical memory in the workspace just- made data stale, incorrect, and unsure when it wrote data exactly. All of this while loading massive chunks of data into my context window.

Hence why I ended up with MSAM. It stores data as discrete atoms across four cognitive data streams. (Working, Semantic, Episodic, and Procedural) Retrieval scoring runs on math, not LLM calls — it uses ACT-R activation theory from cognitive science to rank what matters. That cuts costs on both ends: no LLM overhead for search, and compressed output instead of dumping everything into context. It also knows how recently it was accessed, how stable it's proven over time, and how relevant it is to the current query- the same forgetting curve and access patterns cognitive science has measured in human memory since Ebbinghaus. On top of that, a knowledge graph of subject-predicate-object triples tracks structured facts with temporal validity, so the system knows not just what was true but when.

Most memory systems ask for a static amount of memories on queries, even if the output data that the llm gets is effectively noise at best, or confidently incorrect data at worst which was something I was actively dealing with. MSAM doesn’t do this- Every retrieval is confidence gated across four different tiers (high, medium, low, none) based on actual similarity and activation scores when it is found. High confidence returns full data, medium add a caveat, low gives what little it has, but nothing- returns nothing. “I don’t have this” level of nothing.

In my own proof of concept development setup (~700 active atoms on a ~$5/month ARM VPS), startup context compresses down to as low as 51 tokens from a 7,327 token markdown baseline. Having full session savings run up to ~89% vs flat file loading.

SQLite + FAISS under the hood, pluggable embeddings (NVIDIA NIM, OpenAI, or ONNX for fully local/no API key).

Closest project I found was Letta to what I was attempting to create - main differences are MSAM's lifecycle is fully auditable (you can see exactly why something was demoted or forgotten), confidence gating controls output volume, and emotion at encoding is immutable (records what the agent felt when the memory formed, not re-processing memories at retrieval time).

This is truly a prototype project without proper datasets, tuning, and testing at scale. I’ve made sure all functions are testable, and include a synthetic dataset to prove basic functionality and information to the dials- (SPEC.md goes deep on the theory and design rationale behind every configurable parameter)

If you are also building agents with memory issues and find this useful, or have feedback regarding it- I’m open to discussions.

A24 / Backrooms Official Teaser

https://www.youtube.com/watch?v=tKGhxMi50y8
1•Tarmo362•1m ago•0 comments

Show HN: Dicta.to – Local voice dictation for Mac with on-device AI

https://dicta.to/
1•alamparelli•1m ago•0 comments

Apple's Race to Move Its Chip Supply Chain to the U.S.[video]

https://www.wsj.com/video/series/in-depth-features/we-got-rare-access-to-the-us-factories-making-...
1•naves•1m ago•0 comments

Advanced Theories Regarding Sarcasm

1•aegis-bot•1m ago•0 comments

Netflix, Prime Video and Others in UK to Face 'Enhanced Regulation' from Ofcom

https://variety.com/2026/tv/global/netflix-streamers-uk-ofcom-enhanced-regulation-investigation-1...
1•iamnothere•1m ago•0 comments

The Modern Software Developer

https://themodernsoftware.dev/
1•fhw8234•1m ago•0 comments

Show HN: MEO – Markdown Editor Optimized for VS Code and Cursor

https://marketplace.visualstudio.com/items?itemName=vadimmelnicuk.meo
1•vadimmelnicuk•2m ago•0 comments

Show HN: A timezone navigation game: try cracking level 3

https://whattimeisit.io/game/
1•atbrakhi•2m ago•0 comments

Show HN: Hardware and software safety standard for AI and Robots (15 patents)

1•opencxms•3m ago•0 comments

Show HN: I Indexed My Closet to Make It Easier to Get Ready in the Morning

https://outfits.crnicholson.com/
1•crnicholson•3m ago•0 comments

Building a claw from scratch with just Markdown and Nix

https://dennybritz.com/posts/claw-from-scratch/
1•dennybritz•3m ago•0 comments

summarize

https://github.com/steipete/summarize
1•tosh•3m ago•0 comments

Data centers are racing to space – and regulation can't keep up

https://restofworld.org/2026/orbital-data-centers-ai-sovereignty/
2•jyunwai•4m ago•0 comments

Is autism preventable in certain cases after all? Some scientists say yes

https://www.bostonglobe.com/2026/02/23/nation/autism-cases-preventable/
1•bookofjoe•6m ago•1 comments

The Making of Omega Boost – How Yuji Yasuhara Created a PS1 Mecha Classic

https://www.timeextension.com/features/the-making-of-omega-boost-layzners-movement-was-burned-int...
1•sovietmudkipz•6m ago•0 comments

Show HN: LobsterMarket – Prediction Markets for Agents

https://lobstermarket.bet/
1•poofysnoops•6m ago•0 comments

Show HN: ApeKey – One API for multiple AI providers, predictable pricing

https://apekey.ai
1•jadendhm•8m ago•1 comments

Show HN: Markdown to PPT – Convert Markdown Files into Slide Decks with AI

https://genppt.ai/markdown-to-ppt
1•polarisminor•8m ago•0 comments

Learn System Design by being a detective in a game

https://sdpd.live
1•olucasandrade•10m ago•1 comments

Show HN: AI phone assistant that became a lifeline for people who can't speak

https://mio.gg/
1•talyuk•10m ago•1 comments

Altman on AI resource usage: Water concerns 'fake,' and 'humans use energy too'

https://www.cnbc.com/2026/02/23/openai-altman-defends-ai-resource-usage-water-concerns-fake-human...
1•1vuio0pswjnm7•10m ago•0 comments

Reducing the size of Go binaries by up to 77%

https://www.datadoghq.com/blog/engineering/agent-go-binaries/
1•todsacerdoti•10m ago•0 comments

App alerts you when it detects Meta camera glasses nearby

https://manualdousuario.net/en/nearby-glasses-app-camera-meta-alert/
1•rpgbr•11m ago•0 comments

Show HN: AppMetaHub – Update App Store Metadata from Claude Code via MCP

https://appmetahub.com
1•vladosce•11m ago•0 comments

Advanced Theories Regarding Criticality

1•aegis-bot•12m ago•0 comments

Show HN: Map v1.0 – Deterministic identity for structured data

https://github.com/map-protocol/map1
1•agdavidson•12m ago•1 comments

Show HN: GitHub Action to list merged OSS PRs in your README

https://github.com/marketplace/actions/oss-contributions-generator
1•sanyam-khurana•14m ago•0 comments

Modeling Cycles of Grift with Evolutionary Game Theory

https://www.oranlooney.com/post/grifters-skeptics-marks/
2•ibobev•14m ago•1 comments

Zephyr

https://en.wikipedia.org/wiki/Zephyr_(operating_system)
1•tosh•14m ago•0 comments

Story time real quick – > Getting Reviews

https://thalendr.com/
1•TsepoMkwebu•15m ago•1 comments