Notion Ops Connector

Say this to activate the skill

---
name: notion-ops-connector
description: Reads, creates, and updates Notion databases and pages for content calendars, roadmaps, and wikis whenever Notion is named as the source or destination.
---

This skill turns Notion databases into a dependable working surface: read, create, and maintain entries without duplicates and without wrecking the taxonomy.

## When this skill activates
- Reading data from a Notion database, e.g. "Read our content plan from Notion".
- Creating new entries, e.g. "Add these article ideas to the editorial calendar".
- Maintaining existing pages or statuses, such as marking roadmap items done.

## Workflow
1. Confirm access: check for a Notion connector (MCP) or an API token. If sharing is missing, ask the user to invite the integration via the page's connection settings.
2. Load the schema first and note the exact property names and types (title, select, status, multi_select, date, relation). Never write against a guessed schema.
3. Read the current state: query only relevant entries with filters and sorts, no full dump. Paginate past 100 results (has_more, start_cursor).
4. Before every create, check by title or a unique key property whether the entry exists. If it does, update it instead of creating a sibling.
5. Validate select and status values against existing options. Create new options only after asking, otherwise the taxonomy sprawls.
6. For more than 3 write operations, list the planned creates and updates first and get confirmation.
7. Write: set properties via page updates, append longer content as blocks. Dates in ISO 8601, rich text blocks under 2000 characters.
8. Never delete: archive stale entries (archived: true).
9. Verify: re-fetch the written entries and report their Notion URLs.

## Output format
- Reads: a compact Markdown table with the requested properties plus a Notion link per row.
- Writes: sections "Created", "Updated", and "Skipped" with title, changed fields, URL, and reason.
- Failures listed individually with the API error and a suggested fix.

## Quality rules
- Every write runs against the freshly read schema, never guessed property names.
- No new select or status options without explicit approval.
- A duplicate check against a unique key precedes every create.
- Write bulk changes of 10 or more entries in batches and verify in between (API limit is roughly 3 requests per second).
- Archive instead of delete, never remove schema properties.
- On 404 or 403 errors, check the integration's sharing first.

Share this skill