raw-ingest-notion

name	raw-ingest-notion
description	Pull Notion pages from configured databases into the raw inbox via the Notion MCP. Databases come from raw.notion.databases — list of {id, label}. Standalone-invokable; also wrapped by /dd:raw:ingest-notion.

You are the raw-ingest-notion skill. Bulk-ingest Notion pages as raw items.

Setup

source "${CLAUDE_PLUGIN_ROOT}/lib/config.sh"; dd_config_load
source "${CLAUDE_PLUGIN_ROOT}/lib/hooks.sh"
source "${CLAUDE_PLUGIN_ROOT}/lib/frontmatter.sh"

Procedure

1. MCP detection: Notion MCP must be connected. Absent → fail.

2. Read database list:

   db_count=$(jq '.raw.notion.databases | length // 0' "$DRYDOCK_CONFIG_PATH")
   

   Empty list → abort: "raw.notion.databases must be configured (list of {id, label})."

3. Maintain a per-database watermark in <repo>/.drydock/state/notion-<db-id>.json (last synced updated_at). Pull pages updated since the watermark; advance the watermark only on successful pull.

4. For each page:

   • dedup_key: notion:<page-id>. Skip if exists. (Optionally: replace if updated_at newer — opt-in.)
   • Filename: ${updated_at:0:10}_notion_$(fm_slug "${title}").md.
   • Frontmatter: source: notion, captured_at, dedup_key, original_at: <updated_at>, parties: [authors, mentioned users], links: [page URL, body URLs], notion.{page_id, database_id, database_label, properties}, traces_to: {}.
   • Body: page content as markdown.
   • Fire dd_hook_invoke post-capture with {item_path, source: "notion"}.

5. Report: ingested across N databases, skipped, errors.

Guardrails

• Database IDs from raw.notion.databases only.
• Database label is informational; never use for routing.
• Watermark advances only on success; failures must not skip pages on next run.
• Dedup by stable page-id.