notcrawl

README

🗞️ notcrawl

notcrawl mirrors Notion workspace data into local SQLite and normalized Markdown so you can search, query, diff, and share your Notion memory without depending on the Notion UI.

It has three ingestion paths:

• desktop: read-only snapshots of the local Notion desktop cache
• api: official Notion API sync with rate-limit aware crawling
• notion-mcp: targeted repair through a preconfigured Codex Notion connector

SQLite is the canonical archive. Markdown is the durable human/agent surface. Git share mode publishes normalized snapshots that other machines can subscribe to without holding Notion credentials.

Current Scope

• local SQLite storage with FTS5
• read-only local desktop cache ingestion from macOS Notion
• official API page/block/user/comment ingestion
• Notion database metadata and row ingestion through the official API
• current Notion data-source API support plus legacy database endpoint support
• normalized Markdown export organized by Unicode-safe workspace, teamspace, and page paths
• CSV/TSV export for crawled Notion database rows
• compressed JSONL git-share snapshots plus import/update workflows
• terminal archive browser for quick local page/database inspection
• archive status, activity reporting, and SQLite maintenance commands
• read-only SQL access for ad hoc inspection

Install

brew install openclaw/tap/notcrawl

You can also download archives, .deb, or .rpm packages from the latest release. Official macOS archives are signed with the OpenClaw Foundation Developer ID; local source and snapshot builds remain credential-free.

Check for newer releases manually with:

notcrawl check-update

Interactive terminal runs also perform a cached daily release check and print a stderr notice when a newer OpenClaw release is available. Set NOTCRAWL_NO_UPDATE_CHECK=1 or CRAWLKIT_NO_UPDATE_CHECK=1 to disable that passive notice.

Quick Start

Use the local Notion Desktop cache:

notcrawl init
notcrawl doctor
notcrawl status
notcrawl report
notcrawl sync --source desktop
notcrawl export-md
notcrawl search "launch plan"
notcrawl tui

Or use the official Notion API:

export NOTION_TOKEN="secret_..."
notcrawl sync --source api
notcrawl databases
notcrawl export-db --database DATABASE_ID --format csv --output roadmap.csv
notcrawl export-db --all --dir exports/csv

Or repair incomplete Desktop pages through the Notion app connected in Codex:

notcrawl sync --source notion-mcp
notcrawl sync --source notion-mcp --page PAGE_ID
notcrawl sync --source notion-mcp --query "launch plan" --limit 25

Without --page or --query, this source fetches known Desktop pages that have no cached body or missing referenced blocks, plus API pages whose block sync did not complete. It does not claim to enumerate the entire workspace. The transport reuses Codex authentication and the configured Notion app through the experimental ChatGPT apps gateway. Empty connector bodies leave existing archive content unchanged and remain eligible for a later retry.

Default paths:

• config: ~/.notcrawl/config.toml
• database: ~/.notcrawl/notcrawl.db
• cache: ~/.notcrawl/cache
• Markdown archive: ~/.notcrawl/pages
• git share repo: ~/.notcrawl/share

Commands

• init writes a starter config
• doctor checks config, SQLite, desktop cache, and token presence
• status prints archive counts, last sync time, and database/WAL size
• metadata --json, status --json, and doctor --json expose crawlkit control/status payloads for launchers, automation, and CI
• report summarizes recent page, database, space, and comment activity
• maintain rebuilds FTS, optimizes SQLite indexes, and can run VACUUM
• sync ingests from desktop, api, notion-mcp, or all
• export-md renders normalized Markdown files from SQLite
• databases lists crawled Notion databases
• export-db exports one crawled Notion database, or all databases with --all --dir, to CSV or TSV
• search searches page and comment text through FTS5
• tui opens the terminal archive browser for pages and databases
• sql runs read-only SQL against the archive
• publish exports SQLite tables and Markdown into a git share repo; --tag names an immutable checkpoint
• subscribe clones a share repo and merges the latest snapshot into local rows; --restore opts into exact replacement and --retain-revisions saves replaced local payloads in the local-only record_revisions table
• update merges current or --ref historical share data without changing the checkout; --restore and --retain-revisions select the same retention modes

Shared crawlkit surfaces

notcrawl uses crawlkit for standard config paths, SQLite open/read helpers, snapshot packing/import, git-backed archive sharing, output formatting, status payloads, and the shared terminal explorer. Notion API/Desktop parsing, Markdown rendering, page/comment/database schemas, and Notion FTS bodies remain owned by notcrawl.

The TUI follows the gitcrawl-style three-pane model: workspace/teamspace/page or database groups on the left, pages/databases in the middle, and a readable document preview plus comments and metadata on the right. It supports pane focus, sortable headers, mouse selection, right-click actions, and a local/remote footer.

Distribution

Release packaging runs through the shared OpenClaw GitHub Actions workflow. Tagged releases publish signed and notarized macOS archives, checksums, .deb and .rpm packages, GitHub release notes, and a verified Homebrew tap update.

See docs/distribution.md for release operations.

Safety Model

Desktop mode is read-only. It snapshots Notion's local SQLite database before reading it and never writes to Notion application storage. Desktop cache coverage is opportunistic; Markdown exports mark pages whose referenced blocks were not cached locally, and API sync can fill content shared with an integration. Rows missing from a later Desktop snapshot are preserved because cache eviction is not a deletion signal; explicit Notion tombstones still retire records.

API mode uses the official Notion API. It stores raw API payloads alongside normalized rows so renderers can improve without recrawling.

Notion MCP mode is read-only and targeted. It reads the Codex bearer credential at request time, never stores it, dynamically resolves only the connected Notion search/fetch tools, and strips signed URL credentials before persisting connector Markdown, including page properties. Credentials are sent only to the exact HTTPS ChatGPT apps gateway. The gateway and Codex auth-file format are experimental contracts and may change.

Secrets are never exported into Markdown or git-share snapshots.