npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

midas-memory-mcp

v0.0.4

Published

Midas — local-first, source-traceable agent memory over MCP (TypeScript port, experimental). No LLM at ingest or query.

Readme

midas-memory-mcp (TypeScript) — experimental

A TypeScript port of Midas: local-first, source-traceable memory for AI agents over MCP — no LLM and no network at ingest or query.

Status: experimental. The Python server (pip install "midas-memory[mcp,local]") is the reference implementation with semantic ONNX embeddings, NLI-gated belief revision, and the full eval harness behind it. This port covers the core for Node-first setups.

Why it exists

Most MCP clients live in the Node ecosystem. This package gives them a zero-Python install:

npx midas-memory-mcp        # or: npm i -g midas-memory-mcp && midas-mcp
{
  "mcpServers": {
    "midas": {
      "command": "npx",
      "args": ["-y", "midas-memory-mcp"],
      "env": { "MIDAS_MCP_DB": "/home/you/.midas/memory.sqlite3" }
    }
  }
}

Parity with the Python server

  • Same SQLite schema and float32 blob encoding — a TS server and a Python server can point at the same DB file and share one live memory (both probe SQLite's data_version and refresh on other connections' writes). Verified bidirectionally in tests.
  • Bit-comparable hashing embedder (md5 token hashing, identical sign/index math) — vectors written by one runtime are recalled semantically by the other. Parity is pinned by a fixture generated from the Python implementation.
  • Same tool surfaceremember, capture (policy-gated, no LLM), recall (source-traceable), build_context (lean dated lines + "Today is" anchor), inspect_memory, check_memory_use (provenance guard), forget, forget_matching (dry-run + audit), forget_all, maintain, stats, memory_policy — plus the same env knobs (MIDAS_MCP_DB, MIDAS_MCP_MAX_RECORDS, MIDAS_MCP_MIN_IMPORTANCE, MIDAS_MCP_NAMESPACE, MIDAS_MCP_ACTOR, MIDAS_MCP_SUPERSEDE) and the same injected agent policy text.
  • Same shipping behaviour — relevance × importance × recency ranking, the measured-safe scale-free parsimony floor (minRelevanceRatio 0.3), BM25+RRF hybrid (cached index), typed belief revision with supersession chains, no-LLM importance scoring, selective forgetting with durable-tier protection.

Not ported (yet)

  • Semantic ONNX embeddings (bge / multilingual) — the default embedder here is the offline hashing one (lexical-ish). For real semantic recall today, run the Python server; both can share the same DB.
  • Local NLI contradiction gating, the cross-encoder reranker, and the eval harness.

Requirements

Node >= 22.5 (uses the built-in node:sqlite; you may see its experimental warning on stderr).

npm test   # builds + runs the suite, including the Python-parity fixture

MIT — same license, same repo, same PRIVACY.md posture: everything stays on your machine.