Coda for AI agents
Coda turns documents into flexible tools, combining text, tables, and automation so a team can build trackers and workflows inside a single doc. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.
What agents do in Coda.
Create a new page in a doc
Create a new page inside a doc, ready to hold notes or a table.
Delete a row from a table
Delete a specific row from a table once its data is no longer needed.
Confirm before deleting a doc
Confirm before deleting a doc, since every page and table inside it is removed too.
110 Coda actions, graded by risk.
Every Coda action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
57Look things up. Allowed by default.
- coda.get_a_rowGet a row
- coda.get_a_pageGet a page
- coda.get_folderGet folder
- coda.list_packsList packs
- coda.list_pagesList pages
- coda.get_a_tableGet a table
- coda.list_tablesList tables
- coda.get_a_columnGet a column
- coda.list_columnsList columns
- coda.list_foldersList Folders
- coda.get_a_controlGet a control
- coda.get_a_formulaGet a formula
- coda.get_user_infoGet user info
- coda.list_controlsList controls
- coda.list_formulasList formulas
- coda.list_table_rowsList table rows
write
40Create and change things. Allow, or ask the user first.
- coda.create_docCreate doc
- coda.update_docUpdate doc
- coda.update_rowUpdate Row in Coda Table (Deprecated)
- coda.create_packCreate Pack
- coda.update_packUpdate pack
- coda.create_a_pageCreate a page
- coda.create_folderCreate Folder
- coda.update_a_pageUpdate a page
- coda.update_folderUpdate Folder
- coda.add_permissionAdd permission
- coda.add_custom_domainAdd custom domain
- coda.update_acl_settingsUpdate acl settings
- coda.add_a_maker_for_packAdd a maker for pack
- coda.create_pack_invitationCreate Pack Invitation
- coda.add_a_category_for_packAdd a category for pack
- coda.add_a_permission_for_packAdd a permission for pack
destructive
13Delete, cancel or archive. Ask first, or deny outright.
- coda.delete_docDelete doc
- coda.delete_rowDelete row
- coda.delete_packDelete pack
- coda.delete_a_pageDelete a page
- coda.delete_folderDelete Folder
- coda.delete_permissionDelete permission
- coda.delete_page_contentDelete Page Content
- coda.delete_multiple_rowsDelete multiple rows
- coda.delete_a_maker_for_packDelete a maker for pack
- coda.delete_pack_listing_draftDelete Pack listing draft
- coda.delete_a_category_for_packDelete a category for pack
- coda.delete_a_permission_for_packDelete a permission for pack
- coda.unpublish_docUnpublish doc
Coda in three steps.
- 01Your users connect CodaThey add their Coda api key on Arc0 Connect, under your brand. It goes straight into the vault.
- 02You set the rulesReads run, writes like “create doc” can wait for the user, and “delete doc” can be denied outright.
- 03Any agent can actYour agent calls Coda through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('coda', { read: 'allow', write: 'ask', // create_doc destructive: 'deny', // delete_doc }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How Coda connects.
Users add their Coda api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
The same Coda connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →
- AUTH
- API key
- CREDENTIALS
- Per-tenant encrypted vault
- MODEL SEES
- Results only, never credentials
- AUDIT LOG
- Every call, on every plan
Use Coda from any agent.
Coda and Arc0, answered.
Can I use Coda with Claude, ChatGPT or Cursor?
Yes. Connect Coda to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Coda actions you allow.
How do users connect Coda?
Users add their Coda api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
Which Coda actions can my agent take?
110 in total: 57 read, 40 write and 13 destructive, such as “create doc”. Your policies decide which of them each agent may call.
Can I stop my agent from deleting things in Coda?
Yes. Actions like “delete doc” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.
Can my own backend call Coda too?
Yes. The same Coda connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug Coda into your agent.
Your users connect Coda once, under your brand. Your agent gets 110 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan