I’m still running into a few high-impact gaps where MCP can’t currently do the operation, and those are painful for real workflows (especially automation).
Most impactful to least impactful from what I’m seeing:
- archive/delete pagees
- change page icon
- uploading local files/images (for example
./image.png) breaks unless files are hosted first
- reading very large page properties (eg big relations/people lists) can be incomplete and REST has better endpoints for full/accurate reads
Btw I completely agree with the notion-cli MCP-first where possible over API. That should stay the default. Curious what you would think tho about MCP-first + REST-fallback model
On my fork I have implemented:
I’m still running into a few high-impact gaps where MCP can’t currently do the operation, and those are painful for real workflows (especially automation).
Most impactful to least impactful from what I’m seeing:
./image.png) breaks unless files are hosted firstBtw I completely agree with the
notion-cliMCP-first where possible over API. That should stay the default. Curious what you would think tho about MCP-first + REST-fallback modelOn my fork I have implemented:
auth api setup/status/verify/unset): feat(auth): add official API token setup flow 0xble/notion-cli#10page edit --icon): feat(page): add api-backed --icon support 0xble/notion-cli#15page archivesupport: feat(page): add api fallback for archive command 0xble/notion-cli#11page property getcommand with paginated reads for complete large property values: feat(page): add api command for full property reads 0xble/notion-cli#13