Skip to content

for Zotero · into Obsidian

Your library is the signal.

lit-monitor tracks PubMed, arXiv & Scopus, ranks every new paper by similarity to your Zotero library, and files structured notes into Obsidian — queryable from the terminal, a browser, or any MCP client.

Get started → View on PyPI GitHub
$ pip install lit-monitor
PubMedarXivScopus ObsidianKnowledge graphMCP server

What it does

  • A monitor that runs itself

    Schedule it (launchd / systemd) and new papers arrive already searched, ranked, extracted, and filed — with a dated digest and an OS notification when each run finishes. Open your vault and the work is already done.

  • Library-relative ranking

    Your library is the relevance signal. Each candidate is scored by a transparent six-signal mix — semantic similarity to your Zotero library plus graph signals (shared entities, citation edges, shared authors) — and every paper shows the decomposition, so you see exactly why it ranked where it did.

  • Knowledge graph you can talk to

    A real KuzuDB graph of entities and typed relationships — not just embeddings. Hold a conversation grounded in your corpus from the CLI (ask), an AI client (MCP), or HTTP, with vector, graph, or hybrid retrieval.

  • Obsidian-native output

    Every kept paper becomes a structured Markdown note in your vault — plain .md files you own, with Dataview front matter and persist zones that protect your own annotations across rebuilds. An Obsidian companion plugin is planned.

  • MCP server for AI clients

    Twelve tools that Claude Desktop, Cursor, and any MCP-capable agent can call to query the graph and vector index, including a guarded read-only Cypher escape hatch.

  • Local-first, open, and free

    MIT-licensed and built for researchers: your library is embedded and stored on your machine with no per-call costs and no account — open formats, no lock-in. Cloud providers are strictly opt-in. Contributions welcome.

Quickstart

Ollama is the one external prerequisite (for local embeddings).

pip install lit-monitor        # or: uvx lit-monitor / pipx install lit-monitor
lit-monitor first-run          # interactive setup: credentials + seeds your config
lit-monitor serve              # browse at http://127.0.0.1:8765

New here? Start with the Installation guide, then How it works for the scoring model.

Explore

  • Installation

    Install paths, optional extras, and the prerequisites.

  • Configuration

    Config files, setup recipes, providers, notifications, strict mode, deployment.

  • How it works

    Library-as-signal, the knowledge graph, scheduling, the six ranking signals, and the local-first philosophy.

  • Web UI

    The dashboards and the 8-step setup wizard.

  • MCP server

    Twelve MCP tools for AI clients, each described.

  • HTTP API

    The HTTP query and ingestion surface.

Beta

lit-monitor is feature-complete and in active daily use, but still maturing toward 1.0 — interfaces and on-disk formats may change between releases. Bug reports and feedback are very welcome via GitHub Issues. Provided under the MIT License.