Here’s what we shipped in June 2026.
Happy Pride Month to all who celebrate! We’re also in the midst of some 
World Cup soccer action, Knicks mania in NYC, and feeling like summer has finally arrived in the northern hemisphere 
Lots to toast a 
spritz to this month. But here’s what we’ve been shipping—we’ll keep adding to this post as new things ship, so check back often:
June 10
Product feature: MCP Improvements
What changed: document_read renamed to document_outline
What we did: Now returns structure and metadata only; contentTypesToInclude parameter removed
Why it matters: Outline a doc’s pages and metadata in one fast call, without scanning body content
June 10
Product feature: MCP Improvements
What changed: whoami now returns workspaces
What we did: The whoami tool now also returns the workspaces a user belongs to
Why it matters: Your AI knows which workspaces you’re in, enabling smarter doc creation and search without having to ask
June 10
Product feature: MCP Improvements
What changed: page_read split into two tools
What we did: Split into page_describe (metadata and child outlines) and content_read (body content)
Why it matters: Fetch page metadata without pulling body content, and read content with finer control, including canvas cells inside table rows
June 10
Product feature: MCP Improvements
What changed: New tool: name_match
What we did: Batched fuzzy lookup (up to 20 at a time) of page, table, view, and person names, returning URIs and user profiles
Why it matters: Refer to pages, tables, and people by approximate name and your AI resolves them to the right resource without searching or guessing URIs
June 11
Product Feature: ChatGPT App Directory
What changed: Coda is now an official, verified connector in the ChatGPT App Directory
What we did: Partnered with OpenAI to get Coda listed as a verified connector and reduced setup to a single click. No more pasting tokens, enabling developer mode, or hunting down API URLs
Why it matters: If you’ve wanted to connect Coda to ChatGPT but never got around to the manual setup, that barrier is gone. A few things worth knowing:
- One-click setup: Connect directly from the directory without any configuration steps
- Discoverable by anyone: Coda is now findable by anyone browsing the ChatGPT App Directory, not just people who already know about the MCP
- Already using the Coda MCP? Nothing changed under the hood. This is purely about visibility and easier onboarding for new users
Get started or share it with your team:
Connect via the ChatGPT App Directory
Help Center article
Live webinar June 24
June 16
Product feature: MCP Improvements
What changed: Image and file columns are now writable in table_row_manage
What we did: Publicly hosted imageAttachments, fileAttachments, and render-format image columns are now writable via table_rows_manage
Why it matters: Write images and file URLs straight into table cells, single or multi-value
June 19
Product feature: MCP Improvements
What changed: Admin controls for Personal Access Tokens
What we did: Enterprise admins can now control who in their org can use Personal Access Tokens with MCP, managed directly from Admin settings
Why it matters: More secure MCP usage across your org. Admins can let certain users connect via PAT-based scripts and integrations while blocking others, without affecting OAuth access
June 21
Product feature: MCP Improvements
What changed: Reactions in comment_add and control_manage
What we did: comment_add gains add_reaction and remove_reaction actions; control_manage gains a new reaction control type that takes canvas emoji and icon reactions
Why it matters: Your AI can now place an emoji reaction control on any canvas and add or remove reactions on individual comments, covering the full reaction lifecycle alongside threads and replies
I have a job application doc and I had to do few workarounds before in order able to extract information from the attached pdfs of candidates. I’m going to test it immediately !