BEBERTmackliin·Mar 12, 2026Submission
Anchor Engine – deterministic semantic memory for LLMs, <1GB RAM runs on a phone
1 points · Open discussionrsbalchii
anchor-engine-node
HISTORICAL RECORD2026–2026
rsbalchii / anchor-engine-node · Not publicly available
anchor-engine-node
Discover this repository through its Hacker News conversations.
2Distinct people
7Mentions
0Explicit recommenders
What to know before exploring
What it is
A public description has not been verified.
Where it is used
No documented product use has been verified yet.
What people are saying
No recommendation phrase was found in the indexed comments.
No classified warning is attached to this page. Read the discussions before treating silence as approval.
Also in these discussions
No other repository is repeatedly linked in the same discussions.
This week on Hacker News
Nov 8–Nov 15 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/15/2026, 2:04:58 PM UTC.
No direct repository links were found in the indexed HN discussions over the past seven days. Older discussions are available below.
Mentions over time
HN mentions by 6-hour interval7 dated mentions
Observed mentions: 2026-03-05 21:30:09 UTC to 2026-03-12 14:54:28 UTC.
Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.
View mention counts and date ranges
| Period | Start (UTC) | End (exclusive, UTC) | Mentions | Submissions | Comments |
|---|---|---|---|---|---|
| Mar 5, 18:00 | 2026-03-05 18:00:00 UTC | 2026-03-06 00:00:00 UTC | 1 | 1 | 0 |
| Mar 6, 00:00 | 2026-03-06 00:00:00 UTC | 2026-03-06 06:00:00 UTC | 1 | 0 | 1 |
| Mar 6, 06:00 | 2026-03-06 06:00:00 UTC | 2026-03-06 12:00:00 UTC | 2 | 1 | 1 |
| Mar 6, 12:00 | 2026-03-06 12:00:00 UTC | 2026-03-06 18:00:00 UTC | 1 | 1 | 0 |
| Mar 6, 18:00 | 2026-03-06 18:00:00 UTC | 2026-03-07 00:00:00 UTC | 0 | 0 | 0 |
| Mar 7, 00:00 | 2026-03-07 00:00:00 UTC | 2026-03-07 06:00:00 UTC | 0 | 0 | 0 |
| Mar 7, 06:00 | 2026-03-07 06:00:00 UTC | 2026-03-07 12:00:00 UTC | 0 | 0 | 0 |
| Mar 7, 12:00 | 2026-03-07 12:00:00 UTC | 2026-03-07 18:00:00 UTC | 0 | 0 | 0 |
| Mar 7, 18:00 | 2026-03-07 18:00:00 UTC | 2026-03-08 00:00:00 UTC | 0 | 0 | 0 |
| Mar 8, 00:00 | 2026-03-08 00:00:00 UTC | 2026-03-08 06:00:00 UTC | 0 | 0 | 0 |
| Mar 8, 06:00 | 2026-03-08 06:00:00 UTC | 2026-03-08 12:00:00 UTC | 0 | 0 | 0 |
| Mar 8, 12:00 | 2026-03-08 12:00:00 UTC | 2026-03-08 18:00:00 UTC | 0 | 0 | 0 |
| Mar 8, 18:00 | 2026-03-08 18:00:00 UTC | 2026-03-09 00:00:00 UTC | 0 | 0 | 0 |
| Mar 9, 00:00 | 2026-03-09 00:00:00 UTC | 2026-03-09 06:00:00 UTC | 0 | 0 | 0 |
| Mar 9, 06:00 | 2026-03-09 06:00:00 UTC | 2026-03-09 12:00:00 UTC | 0 | 0 | 0 |
| Mar 9, 12:00 | 2026-03-09 12:00:00 UTC | 2026-03-09 18:00:00 UTC | 0 | 0 | 0 |
| Mar 9, 18:00 | 2026-03-09 18:00:00 UTC | 2026-03-10 00:00:00 UTC | 0 | 0 | 0 |
| Mar 10, 00:00 | 2026-03-10 00:00:00 UTC | 2026-03-10 06:00:00 UTC | 0 | 0 | 0 |
| Mar 10, 06:00 | 2026-03-10 06:00:00 UTC | 2026-03-10 12:00:00 UTC | 0 | 0 | 0 |
| Mar 10, 12:00 | 2026-03-10 12:00:00 UTC | 2026-03-10 18:00:00 UTC | 0 | 0 | 0 |
| Mar 10, 18:00 | 2026-03-10 18:00:00 UTC | 2026-03-11 00:00:00 UTC | 0 | 0 | 0 |
| Mar 11, 00:00 | 2026-03-11 00:00:00 UTC | 2026-03-11 06:00:00 UTC | 0 | 0 | 0 |
| Mar 11, 06:00 | 2026-03-11 06:00:00 UTC | 2026-03-11 12:00:00 UTC | 0 | 0 | 0 |
| Mar 11, 12:00 | 2026-03-11 12:00:00 UTC | 2026-03-11 18:00:00 UTC | 0 | 0 | 0 |
| Mar 11, 18:00 | 2026-03-11 18:00:00 UTC | 2026-03-12 00:00:00 UTC | 0 | 0 | 0 |
| Mar 12, 00:00 | 2026-03-12 00:00:00 UTC | 2026-03-12 06:00:00 UTC | 0 | 0 | 0 |
| Mar 12, 06:00 | 2026-03-12 06:00:00 UTC | 2026-03-12 12:00:00 UTC | 0 | 0 | 0 |
| Mar 12, 12:00 | 2026-03-12 12:00:00 UTC | 2026-03-12 18:00:00 UTC | 2 | 1 | 1 |
What Hacker News says
2 sourcesI built Anchor because I kept hitting the same wall: local LLMs are great, but every conversation is a fresh start. Vector search is the default hammer, but for structured memory—project decisions, entity relationships, temporal facts—it's often the wrong tool.
Live demo (in-browser, no setup):
https://rsbalchii.github.io/anchor-engine-node/demo/index.ht...
Search Moby Dick or Frankenstein and see the tag-based receipts that show why each result matched.
How it works
Anchor uses graph traversal (the STAR algorithm) instead of embeddings. Concepts become nodes, relationships become edges. The database stores only pointers (file paths + byte offsets); content stays on disk, so the index is small and rebuildable. PGlite (PostgreSQL in WASM) lets it run anywhere Node.js does – including a Pixel 7 in Termux, with <1GB RAM.
Performance
- <200ms
Read the rest
p95 search on a 28M-token corpus
- <1GB RAM – runs on a $200 mini PC, a Raspberry Pi, or a phone
- Pure JS/TS, compiled to WASM, no cloud dependencies
What’s new in v4.6
- distill: lossless compression of your corpus into a single deduplicated YAML file. I tested it on 8 months of my own chat logs: 2336 → 1268 unique lines, 1.84:1 compression, 5 minutes on a Pixel 7.
- MCP server (v4.7.0) – exposes search and distillation to any MCP client (Claude Code, Cursor, Qwen tools)
- Adaptive concurrency – automatic switching between sequential (mobile) and parallel (desktop) processing
The recursion
I used Anchor to build itself. Every bug fix and design decision is in the graph – that's how I kept the complexity manageable.
Where it fits
If you're building local agents, personal knowledge bases, or mobile assistants and want memory that's inspectable, deterministic, and lightweight – this is for you.
GitHub repo:
https://github.com/RSBalchII/anchor-engine-node
Whitepaper:
https://github.com/RSBalchII/anchor-engine-node/blob/main/do...
Happy to answer questions about the algorithm, the recursion, or the mobile optimizations.