notcrawl

Why it's ranked here

This is a serious archival tool, not merely a content exporter. Read-only Desktop ingestion, raw API payload retention, incomplete-content warnings, read-only SQL, maintenance commands, and merge-aware sharing show careful treatment of long-lived data. Its main reservation is coverage: the fastest source only sees locally cached content, while the connector remains experimental and targeted.

What's good

The source choices solve distinct problems without pretending they are equivalent. Desktop mode needs no token and snapshots the cache before reading. API mode captures shared pages, blocks, users, comments, databases, and rows. Missing Desktop blocks stay visible as warnings and repair candidates. Git snapshots exclude secrets and preserve local-only rows during normal merges.

Tradeoffs

Desktop coverage is opportunistic, so an uncached page body may remain incomplete until another source fills it. API discovery depends on integration sharing and token scope. The MCP route cannot enumerate a whole workspace, uses experimental gateway and credential-file contracts, and requires Codex configuration. Database row ingestion does not fetch each row's blocks or comments. Published support names macOS and Linux, not Windows.

How to use it well

Use it when engineers, researchers, or agents need repeatable local search, historical diffs, database exports, or credential-free archive sharing. Start with Desktop for speed, add the API for authoritative shared content, and reserve MCP for bounded repairs. It does not replace Notion editing, full connector enumeration, or guaranteed capture of uncached private material.

Observed

License

MIT

Primary language

Go

Install surface

Homebrew, macOS and Linux archives, Debian packages, and RPM packages

Interfaces

Command-line interface, three-pane terminal browser, JSON automation output, and read-only SQL

Supported platforms

macOS and Linux

Storage

SQLite with FTS5, normalized Markdown, and gzip-compressed JSONL Git snapshots