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

@theglitchking/semantic-memory

v1.2.0

Published

Unified memory-layer plugin for Claude Code. One MCP server, multiple typed corpora (vault, code, plans, docs, research, project-map), unified cross-corpus knowledge graph, conditional tool registration, in-process drift detection, and human↔code translat

Readme

semantic-memory (formerly semantic-sidekick)

A shelf is retrievable. A prosthesis is present. A vault is local. Memory is layered.

v1.0.0 rebrandsemantic-sidekicksemantic-memory. The npm package, plugin name, MCP server name, and marketplace identity all change. Internal storage layout (.semantic-sidekick-index/, ~/.semantic-sidekick/models/) is preserved — existing installs continue to read their indices and model caches without re-indexing or re-downloading. See CHANGELOG.md for the full migration. The bin/semantic-sidekick command remains as an alias for bin/semantic-memory during the transition.

semantic-memory is the unified memory-layer plugin for Claude Code. One MCP server, multiple typed corpora (vault, code, plans, docs, research, project-map), a unified cross-corpus knowledge graph, conditional tool registration based on which corpora are active, in-process drift detection, and human↔code translation via the babel-fish glossary side-channel.

The original semantic-sidekick framing — activation / mode-routing / capture layer on top of semantic-pages — is preserved as the vault corpus's behavior. What's new in 1.0.0 is the multi-corpus architecture that absorbs documentation, code, plans, research, and babel-fish project-maps into the same MCP surface.

Status

Phases 1–4 shipped (2026-04-22). Layers 2–5 fully implemented and committed to feat/sidekick-layers. Phase 5 (advanced calibration + contradiction detection) is explicitly gated per the plan ("only build when Phase 4 has been in use for ≥2 weeks and a specific failure demands it").

| Phase | What it adds | Commit | |---|---|---| | 1 — Activation | vault-first skill + SessionStart/UserPromptSubmit hooks + CLAUDE.md cite-or-deflect rule | c8f722f | | 2 — Structure + Capture | schema, apply_patch, synthesize_note, lint (schema/provenance/stale), Stop-hook capture-on-close, parsed-doc cache (9x hook speedup) | 1807c58 | | 3 — Ingest + Maintenance | ingest_source, structured log.md + log_event/log_query, hierarchical index auto-regen, broken-link lint | b8d4492 | | 4 — Routing + Transitions | research-mode + outage-silence skills, /mode + /vault commands, mode-aware hooks, transition-capture prompts | 75ee813 |

Per-phase design docs live in .documentation/phase-{1..4}-*.md.

Quickstart as a Claude Code plugin

# Add the marketplace (from git)
/plugin marketplace add https://github.com/TheGlitchKing/semantic-sidekick

# Install the plugin
/plugin install semantic-sidekick

On install:

  • .mcp.json is reconciled to point at .claude/.vault (or your --notes <path>).
  • Skills (vault-first, research-mode, outage-silence) are linked into .claude/skills/.
  • Hooks (SessionStart, UserPromptSubmit, Stop) register from hooks/hooks.json.
  • Commands (/mode, /vault) become available.

First run: semantic-sidekick --notes ./path/to/vault --reindex to build the vector index (writes .semantic-sidekick-index/). Subsequent hook calls hit the cache in <1s.

Using the modes

| Command | Effect | |---|---| | /mode research | Enter research-mode. Every source introduced gets filed via ingest_source; mandatory synthesize_note on exit. Visible [research] prefix. | | /mode outage-silence | Enter outage-silence. No proactive vault search. Terse responses. Postmortem drafted as synthesize_note on exit. Visible [outage] prefix. | | /mode vault-first | Return to default. Project-scoped prose questions trigger vault consultation with cite-or-deflect. | | /mode status | Print the current mode. | | /vault <query> | Explicit vault query. Always works, even in outage-silence. |

33 MCP tools shipped

Search (4): search_semantic, search_text, search_graph, search_hybrid
Read (3): read_note, read_multiple_notes, list_notes
Write (4): create_note, update_note, delete_note, move_note
Metadata (4): get_frontmatter, update_frontmatter, manage_tags, rename_tag
Graph (4): backlinks, forwardlinks, graph_path, graph_statistics
System (2): get_stats, reindex
Patch/Synthesis (4): apply_patch, synthesize_note, ingest_source, install_schema
Lint (4): find_schema_violations, find_missing_provenance, find_stale, find_broken_links, lint_vault
Log/Maintenance (3): log_event, log_query, regenerate_index

Relationship to semantic-pages

  • Substrate (retrieval, graph, MCP tools): inherited unchanged from semantic-pages.
  • Upstream remote tracks semantic-pages for cherry-picking substrate fixes. Push is disabled; sidekick does not push back.
  • What's net-new in sidekick: activation hooks, mode-routing skills, capture workflows (ingest/synthesize), lint suite, provenance + schema layer.

Architecture (five layers)

  1. Substrate — markdown vault + hybrid search + graph + MCP (inherited)
  2. Structure — schema, provenance frontmatter, atomic apply_patch, structured log
  3. Workflowsingest_source, synthesize_note, lint suite
  4. Activation — SessionStart / UserPromptSubmit / Stop hooks, vault-first skill, CLAUDE.md rules
  5. Routing — mode skills (research / default / outage), signal weighting, transition capture

See the originating design discussion in .planning/semantic-sidekick/ for rationale and phase-by-phase build plan.

Development

npm install
npm run build
node dist/cli/index.js --version

License

MIT