Nov 19, 2026: MCP updates
Named MCP connectors for vendor-hosted MCP servers
You can now pick a vendor's MCP server by name when you create an MCP connection. A new MCP servers group in the application list holds the vendors Celigo has verified, and each one opens a Simple form that asks only for what that vendor needs: a name and Save & authorize for vendors that support automatic client registration, or your own app's client ID and secret with everything else prefilled and the vendor's setup steps shown before you authorize. The full MCP form is one toggle away, pasting a known vendor's URL into the generic MCP connection offers the named connector, and manual URL entry still works. Previously, connecting to a vendor's native MCP server meant finding its endpoint, auth model, app type, and scopes yourself. See Create an MCP connection.
Request status, password reset, and Google re-authentication on the MCP server registry portal
After you request access to an MCP server on the registry portal, the server card and server page now show whether your request is pending, approved, or declined, with the decision date; for an approved service token, your administrator shares the token with you outside the portal. A password reset requested from the portal now emails a link back to the portal, where you finish the reset, and members who sign in only with Google can reset MFA, revoke trusted devices, and view or regenerate the MFA secret by confirming with Google. The Deny link in the access request email now asks the approver to confirm before the request is denied. Previously, a request ended at "request sent", and two sign-in paths sent portal-only members to a workspace they don't have. Also fixed from launch: approval drawer copy and placeholder text, the accessibility role of the Make the registry public switch, and the API response when an approval doesn't match the request. See Publish MCP servers to your registry portal and Use the MCP server registry portal.
URL-first MCP connections — checks before you save, and clearer failures
When you change an MCP connection's URL, or switch an existing connection to Automatic client registration, Celigo now checks the server's authentication settings before anything is saved: a new connection rediscovers the server every time its URL changes, and on an existing connection a pending switch returns to Pre-registered client with the reason if the server can't support automatic registration, keeping your saved OAuth 2.0 client and scopes. When automatic registration fails at authorization, the connection status banner shows the full reason, the vendor's response, and View log, and you switch to a pre-registered client in the Client registration field with the scopes Celigo found filled in; a new OAuth 2.0 client opened from the connection is prefilled down to the domain names and token settings, leaving only the client ID and secret. Previously, a connection learned what a new server supports only when you selected Save & authorize, after the save had already cleared the OAuth 2.0 client and endpoints you had working. See Create an MCP connection.
Nov 12, 2026: Knowledge bases in AI studio and Agent Builder, Microsoft Fabric Warehouse support, new data ingestion targets, MCP updates
Knowledge bases in AI studio and Agent Builder
You can now create and manage knowledge bases in integrator.io under AI studio > Knowledge bases and attach them to AI agents in each agent's settings. Add PDF files from Celigo Storage or upload new ones, set how old content can get before agents are warned or blocked, see which knowledge bases need attention, and review each one's audit log. Previously, creating a knowledge base and attaching it to an agent took API requests. See Create and manage knowledge bases for AI agents and Create an AI agent.
MCP server registry portal and publishing controls
Your MCP servers now have a registry portal: one page where people sign in with their Celigo end-user identity to see the MCP servers they can use, each one's connection URL and capabilities, and the others they can request. If you make the registry public, anyone with the link can browse the servers you list. You choose what's listed with a Public registry switch on each enabled MCP server and Make the registry public in Company > AI settings, both off by default, and access requests arrive on a new Approvals tab beside End users, where approving grants the server with the capabilities you choose or issues a server-scoped API token for a client that can't sign in. Previously, sharing MCP servers meant sending individual URLs around and turning requests into end-user invites by hand. Users with Manage access to MCP servers can no longer create, edit, or delete them; owners and administrators can. See Publish MCP servers to your registry portal and Use the MCP server registry portal.
End-user connections for upstream MCP servers
When a Celigo MCP server proxies a third-party MCP server, each end user can now connect with their own credentials instead of sharing a service account, so their calls run with their own permissions in the upstream application and its audit trail names them. You can require end-user authentication per upstream connection or per proxied tool; a user who hasn't connected is prompted to authenticate rather than falling back to the shared identity, and API-token, personal-access-token, and workspace sign-in callers keep using the shared identity. See Configure end user credentials for an MCP server connection and Understand and add MCP proxies.
URL-first setup for MCP connections
You can now connect to a remote MCP server by entering its URL. Celigo discovers the vendor's OAuth endpoints and scopes from the server, and a new Client registration choice lets Celigo register itself automatically with vendors that support Client ID Metadata Documents, or prefills a pre-registered OAuth 2.0 client so you enter only the client ID and secret. Previously, you read the vendor's OAuth documentation, registered an app in their developer portal, and typed the grant type, token URLs, and scopes by hand. See Create an MCP connection from a server URL.
Look up an end user by email in the End Users API
GET/v1/endusers now accepts an optional email query parameter that returns the end-user record for that address in your account, so a flow or script that syncs MCP end users from an HR system or identity provider can look one user up directly. The match is exact, case-insensitive, and ignores surrounding whitespace, and it combines with the existing provisionedBy filter; an unknown email returns 200 with an empty list, and a malformed email returns 400. Previously, the End Users API filtered only by how a user was provisioned, so finding one user meant fetching the whole list, and a large account could hit the list limit first. Access is unchanged: owners and administrators only.
Guided start options when creating an MCP server
Selecting Create on the MCP servers page now offers a choice of starting points instead of opening an empty configuration form: build the server manually, install a prebuilt MCP server from the Marketplace and customize it, or describe what you need and have Ora propose a server assembled from your existing account resources. Previously, a first MCP server meant a blank form and already knowing which connections, flows, APIs, and Tools were worth exposing. Existing MCP servers are unchanged, and every existing way of creating one still works. See Choose how to create an MCP server.
Bound environment shown on the MCP consent page
When you connect an AI client such as Claude, Cursor, or ChatGPT to the Celigo Platform MCP server through an environment-specific URL (/celigo-mcp/{accountAlias}), the consent page now shows the environment the grant applies to as a read-only label, so you can see which environment the client will work in before you approve access. Previously, an environment-specific URL locked the consent page to one environment but hid the environment name along with the picker. The unaliased /celigo-mcp URL keeps its environment picker and is unchanged.
Connectivity - Microsoft Fabric Warehouse connector
You can now load data into Microsoft Fabric Warehouse by selecting a destination table, the same way you do for Snowflake and SQL Server, and Celigo handles staging and loading. The connector adds SQL exports, Microsoft Entra ID service principal authentication with a client secret or a certificate, the Use optimized bulk load (recommended for larger imports) query type as the default for new imports, optional primary keys, and a Load data as available setting. Previously, reaching Fabric took an HTTP, file, and SQL connector workaround; the prebuilt connector keeps working and is now listed as Microsoft Fabric (REST API). For detailed information, see Set up a connection to Microsoft Fabric Warehouse.
Data ingestion - Microsoft Fabric Warehouse as a Data Ingestion sync target
You can now select Microsoft Fabric Warehouse as a managed Data Ingestion sync destination, alongside Snowflake, BigQuery, NSAW, and Amazon Redshift. Celigo creates the destination schema and table on the first sync, applies schema drift between runs, and upserts when a primary key is configured, loading through a staging folder in your own Fabric Lakehouse. Previously, Fabric customers had no managed ingestion path and built custom pipelines or used external tools. For detailed information, see Create a sync to Microsoft Fabric Warehouse.
Data ingestion - Schema drift notifications
You can now get an email when the schema of a table in a Data Ingestion sync changes, so you hear about added or removed columns and tables without opening Event Logs. A Schema drift alerts option sits next to Sync error alerts on the integration Notifications tab and in sync settings; subscribe to one sync or to all syncs in an integration, and each email lists up to 25 changes, newest first, with a link to Event Logs for the rest. Previously, drift was recorded only in Event Logs, and the only subscription available covered sync errors. For detailed information, see Edit or view a sync's settings.
Data ingestion - Azure Synapse Analytics as a Data Ingestion sync target
You can now select Azure Synapse Analytics (dedicated SQL pool) as a managed Data Ingestion sync destination, alongside Snowflake, BigQuery, NSAW, Amazon Redshift, and Microsoft Fabric Warehouse. Celigo creates the destination schema and table on the first sync, applies schema drift between runs, and upserts when a primary key is configured, loading through COPY INTO from your own Azure Data Lake Storage Gen2 container. Previously, syncing into Synapse meant row-by-row SQL inserts with no auto-DDL or drift handling. For detailed information, see Create a sync to Microsoft Azure Synapse Analytics.
Data ingestion - Trigger a flow when a Data Ingestion sync completes
You can now configure a Data Ingestion sync to run one or more flows automatically when a sync run completes, scheduled or manual, through a new Next integration flow setting, so downstream transformation jobs such as Snowflake tasks or dbt Cloud runs no longer rely on fixed-offset scheduling. The triggered flow's run history shows which sync started it. Existing syncs are unaffected; the setting defaults to empty. For detailed information, see Edit or view a sync's settings.
B2B Manager - EDI Dashboard session persistence and grouped shared transactions
The EDI Dashboard's Lifecycle View groups related EDI documents by PO number or another shared identifier, adds unified search and an expandable lifecycle tab ordered newest to oldest, color-codes Functional Acknowledgment statuses, lets you switch between the Lifecycle and Transaction views, and saves custom dashboard views for reuse. See Track an order's EDI documents with the dashboard's lifecycle view.
Integration apps - Clearer PGP signature errors in Vendor Payment Manager
When a bank file fails PGP signature verification in the Vendor Payment Manager – NetSuite integration app, the error now names the actual cause. Files signed with an insecure hash algorithm (SHA-1, MD5, or RIPEMD-160), files signed with an expired, revoked, or not-yet-valid key, and outgoing files encrypted with an expired bank public key each get their own message telling you what to ask your bank for. Previously, a SHA-1-signed file failed with “Public Key does not match the Private Key that was used to sign the file,” which sent teams to check their keys instead of the bank’s hash algorithm. Signed bank files still decrypt when the bank signs with SHA-256 or stronger; SHA-1, MD5, and RIPEMD-160 signatures remain unsupported. See Configure PGP encryption for Vendor Payment Manager, Bank to NetSuite file transfer settings., and NetSuite to Bank file transfer settings.
Nov 3, 2026: Integration apps and templates
Email remittances matched to bank file payments in Cash Application Manager
In the Cash Application Manager (CAM) for NetSuite integration app, Email Remittance now uses a customer’s remittance email to find the invoices for a payment that’s already in your bank file. It no longer creates a separate payment from the email. Previously, a payment that also arrived in the bank file could post twice. Automated clearing house and wire payments without invoice numbers stayed unapplied until someone found the email. Now CAM finds the open bank payment that matches each email by amount, date, and currency. It adds the email’s invoices and applies the payment, so each payment posts once. Emails to review appear under a new Email remittances view on the Cash Application Manager dashboard in NetSuite. There, your team can approve a match, link the right bank payment, retry matching, or ignore the email. Each mailbox now uses one flow instead of two. In NetSuite sandbox accounts, update the CAM bundle yourself. If you used an invoice prefix or identifier on an email flow, set it on the bank flow. See Review email remittances on the Cash Application Manager dashboard.
November 2, 2026: Knowledge bases for AI agents, Topics replay and execution logs, APIM updates, Celigo File storage connector
Note: These features will roll out throughout the week of November 2, 2026.
Topics replay, execution logs, and lifecycle carry
Rewind a flow's topic subscriptions from the flow builder to redeliver messages it already consumed, starting from the earliest retained message or a specific time, with an optional end time. Previously an outage, a bug, or a mapping mistake meant messages were gone once a flow read past them. A replay runs as its own job delivering only to the requesting flow, in parallel with live delivery and without touching the live cursor, so a listener shared across flows never rewinds. Topic publishes and deliveries now appear in execution logs under your existing log preferences — delivery entries carry the topic name, poll timestamp, and page offsets, publish entries carry the topic name, batch timestamp, and status code, and Detailed or Debug adds payloads on both sides. Cloning an integration, exporting a template, promoting between environments, and ILM pulls carry topic configuration into the target, while retained data, counters, and cursor state never transfer. Existing listeners and publish steps are unchanged. Learn more about Topics.
API management - Schema rules and optional validation for API builder
You can now mark API request and response fields as required, limit them to a list of values, and set a pattern, a length, a numeric range, and a description, field by field or from a JSON schema; those rules appear in the OpenAPI specification. Previously, none of these constraints were enforced or published, so callers could omit a field or send a value outside the range you intended, and the usual fix was a script. Now you turn on Enable request/response validation per API when you're ready: an invalid request returns HTTP 400 and the API doesn't run, a response that breaks the schema returns HTTP 500, and required query parameters are checked for presence. Existing APIs keep working with validation off. Learn how to create and manage APIs using API builder.
Celigo File storage connector
Celigo Storage now appears as its own connector, so builders find it in the connections list alongside S3, FTP, and Drive instead of reaching it through the Celigo APIs connector. The connection form matches Celigo APIs — including file storage path scoping, PGP encryption, and concurrency settings — minus the custom API and listener selections that don't apply. Transfer files out, transfer files in, and look up additional files behave as they do on Celigo APIs, and record-based export, import, and lookup steps expose the File Storage resource operations: list, search, get metadata, delete, create and delete folder, move, and rename. Upload and download stay on the file transfer steps rather than the record-based ones. Learn how to set up a connection to Celigo file storage.
Connectivity - Execution logs for Salesforce in API Builder and Tool Builder
Salesforce steps running inside an API or a Tool now produce execution logs, where previously that detail existed only in Flow Builder and you couldn’t see what was sent to or returned from Salesforce when troubleshooting an API invoke or tool run. Depth follows the API log level in effect: Basic records step metadata on success and adds request and response payloads with the Salesforce error when a step fails, while Standard stores payloads for every call, success and error alike. Logs appear in the existing API Builder request history and Tool Builder debug panels, with Salesforce rendering like an HTTP step, and payload shape matches what Flow Builder already captures for the same operation. Credentials, OAuth tokens, session IDs, and refresh tokens are never logged at any level. Learn how to view API request and execution logs in API builder.
Per-environment MCP server URLs
Each integrator.io environment now has its own Celigo MCP server URL — your region’s Platform MCP endpoint followed by the environment’s MCP account alias, in the form https://api.integrator.io/celigo-mcp/{accountAlias} (for example, https://api.integrator.io/celigo-mcp/acme-corp-sandbox) — so MCP clients such as Claude and Claude Code can stay connected to production and sandbox at the same time. Previously, one connection meant one environment chosen at consent, and switching meant disconnecting and reconnecting. Now consent for a per-environment URL is locked to that environment, each connection has its own token limited to that environment, and the existing unaliased URL, https://api.integrator.io/celigo-mcp, keeps working with the environment picker. For the EU, AU, and CA regions, use api.eu.integrator.io, api.au.integrator.io, or api.ca.integrator.io in place of api.integrator.io.
Connectivity - Optimized bulk load for Azure Synapse Analytics imports
Synapse imports now load through the pool’s native COPY INTO path instead of row-by-row and batched SQL inserts, so multi-million-row imports finish in a fraction of the time. integrator.io stages records in your own ADLS Gen2 container, so data stays in your Azure tenant. A new Use optimized bulk load (recommended for larger imports) query type is the default for new imports on connections that support it, with optional primary keys for upsert, and the existing Use bulk insert SQL type is renamed Use batch insert SQL, unchanged. One-time setup is required: a service principal connection, Storage Blob Data Contributor on the container, Mapper 2.0, and a dedicated SQL pool. Existing imports are unaffected.
Connectivity - OAuth 2.0 for NetSuite connections
You can now authenticate a NetSuite connection with OAuth 2.0 (Client Credentials). Previously, NetSuite connections supported only token-based authentication (TBA), which NetSuite blocks for new integrations from its 2027.1 release; now Celigo creates and renews the certificate and mints tokens for you, an existing connection or installed integration app can switch from TBA to OAuth 2.0 in place without reinstalling or relinking flows, and both TBA options are labeled as deprecating when you create a new connection. Existing TBA connections keep working with no end-of-life date, and a NetSuite admin must first turn on the OAuth 2.0 feature under Setup > Company > Enable Features > SuiteCloud.
Connectivity - Flexible handlebars templates for JSON and XML file generation
File-based imports no longer require a specific template structure when generating JSON or XML. Write handlebars templates using any combination of helpers and expressions: {{#each batch_of_records}} and explicit header and footer structure are no longer required, though a non-blocking warning appears when each block is absent. Preview now validates JSON well-formedness. Learn about generating JSON and XML import files with handlebars.
Array index access in Handlebars helpers
Pass a single array element to a Handlebars helper by adding its index to the path, as in {{{jsonSerialize record.message.blocks.0}}}. Previously such a path failed when passed to a helper, so picking one element out of an array and then serializing, filtering, or summing it meant extra steps or intermediate fields — worse with nested arrays. Indexes work at any depth, for example record.orders.0.lines.1. Existing templates are unchanged.
Connectivity - Default file name for new file imports
New file imports now default the File name to file-{{timestamp "x"}}.<ext>, which writes epoch milliseconds and avoids characters such as the colon that Celigo file storage and Azure Data Lake reject. Previously, the default {{timestamp}} format produced a name some destinations refused with an unclear error. Now the default is safe on every file provider, existing imports are not changed, and an invalid file name returns an error that names the File name field and lists the characters that are not allowed. Learn how to import or transfer files to an FTP server.
Connectivity - Paging stop conditions for HTTP exports
HTTP and REST exports that use the Custom request body paging method now stop when a boolean in the response matches the Paging complete values you enter, and case or whitespace differences no longer prevent a match. Previously, a real JSON boolean such as hasMore: false never matched the string false stored on the export, so the export kept requesting pages after the API had signaled the last page. Records on the last page are still included. For the Page number parameter and Next page token methods, Path to paging complete field in HTTP response body is still checked only when Override HTTP status code for last page is set; set it to 200 to use a body-based stop condition with those methods. Exports that were paging past a boolean or case-variant stop value now stop earlier, so review any whose downstream steps depended on those extra pages.
MCP end-user group sync and capability set improvements
Group sync now resolves end-user group membership from Microsoft Entra ID, Okta, OneLogin, JumpCloud, Duo, and Auth0, where enabling it previously produced no membership and no error; unmatched provider groups are created at sign-in without capability grants. Google Workspace can't send group membership over OIDC, so invite those end users directly. Deleting a capability set now converts access to direct grants rather than stripping it silently, capability sets accept prompts and resources, and you can set capabilities from the MCP server's Access tab. See Set up authentication for an MCP server and Create and manage capability sets on MCP servers.
Account details in error notification emails
Flow error notification, error assigned, and error unassigned emails now end with the Account details table (Region, Company with the account owner's email, and Environment), so you can tell which account an alert is about without signing in. Learn about setting up email error alerts or notifications.
Connectivity - Security warning for HTTP (non-HTTPS) endpoints
integrator.io now shows a warning when a connection is configured with an unencrypted http:// endpoint, in the connection form and when a connection is created or updated through the platform API. Previously, plain HTTP endpoints were accepted with no signal that traffic is unencrypted. The warning is informational: https:// endpoints are unaffected, existing HTTP connections keep working, and switching the endpoint to HTTPS removes it.
October 29, 2026: Integration apps and templates
P2R NetSuite integration app SuiteApp migration
The Payout to Reconciliation Automation for NetSuite integration app moves from bundle-based distribution to a NetSuite SuiteApp, meeting NetSuite's current distribution requirement and simplifying version upgrades. New customers onboard through the NetSuite SuiteApp Control Center instead of the bundle, and Celigo migrates existing bundle customers. See: Install the Payout-to-Reconciliation Automation for NetSuite app.
AI assist for Shopify - NetSuite integration app
The Shopify-NetSuite integration app now includes Celigo AI assist, an add-on that uses Celigo AI to explain why Shopify orders didn’t sync to NetSuite and what happened to an order, refund, or return. You no longer need to review each error or open Shopify, NetSuite, and integrator.io separately. The add-on adds two flows for each store. The first summarizes up to five Shopify orders, refunds, or NetSuite RMAs at a time, with the current status, gaps between Shopify and NetSuite, the next owner, and the next action. The second groups failed order syncs into error patterns, each with its cause, severity, owning team, revenue at risk, and a retry checklist. You review the results and request summaries on a new Celigo AI assist dashboard in NetSuite. AI assist is read-only: it doesn’t change Shopify data, NetSuite transactions, or integrator.io errors. It uses Celigo AI credits. Celigo installs the add-on on existing integration apps, and you can uninstall it and install it again later. In NetSuite sandbox accounts, update the integration bundle yourself. See Use Celigo AI assist for Shopify - NetSuite orders.
AI assist for Amazon- NetSuite integration app
The Amazon - NetSuite integration app now includes AI assist, a free add-on that uses Celigo AI to forecast inventory, flag FBA inventory exceptions, and investigate settlements that don’t reconcile. You no longer need to review each SKU or settlement line yourself. The add-on adds two flows for each store. The first forecasts when each FBA and merchant-fulfilled (MFN) SKU will run out, calls out SKUs that an upcoming sales event such as Prime Day or Black Friday puts at risk, and flags FBA exceptions such as stockouts and unfulfillable units with a likely cause and recommended checks. The second diagnoses the root cause of each settlement variance, explains the Amazon fees involved, and lists the steps to fix it, in order. You review the results on a new Celigo AI assist - Amazon dashboard in NetSuite, which has Inventory forecast and Settlements views. AI assist is read-only: it never writes inventory quantities, listings, or financial records. It uses Celigo AI credits. Celigo installs the add-on on existing integration apps, and you can uninstall it and install it again later. In NetSuite sandbox accounts, update the integration bundle yourself. See Use Celigo AI assist for Amazon - NetSuite inventory and settlements.
October 27, 2026: Knowledge bases for AI agents, on-premise agent groups, EDI Dashboard, MCP management, Celigo Storage activity log
Knowledge bases for AI agents
Build a knowledge base from files already in Celigo File Storage and attach it to your AI agents, so they answer from your own policies, documentation, and contracts rather than the model’s general knowledge. Previously, grounding an agent in company content meant building and maintaining retrieval tooling outside Celigo. Create a knowledge base under AI studio by picking existing files or uploading new ones — files are referenced, not copied — and Celigo indexes them in the background, re-indexes what changes, and stops retrieving a removed file within five minutes. Set how old content can get and whether stale content is used with a warning or blocked, with the knowledge base list flagging anything that needs attention. Attach one or more knowledge bases to an agent in Agent Builder, and preview what the agent retrieved and from which files before publishing. Every run records which knowledge base was used, how old the content was, whether retrieval was allowed, warned, or blocked, and what came back. A knowledge base is registered to an integration, so Manage access governs who can change it and Monitor access who can view it. Existing agents are unchanged until you attach one. See Create and manage knowledge bases for AI agents.
Connectivity - On-premise agent groups (high availability)
You can now run several on-premise agents as a group and point a connection at the group instead of a single agent. New work moves to healthy members when one host is unavailable, you can drain a member to patch its host, and members update one at a time. High availability needs at least two members on separate hosts running agent 3.0.7 or later. Learn more about on-premise agents.
B2B manager - EDI Dashboard lifecycle view
A new Lifecycle View groups every EDI document tied to a single order — purchase order, functional acknowledgment, advance ship notice, invoice — into one searchable timeline. Previously, tracing an order across its full EDI trail meant correlating transaction records scattered across the dashboard by hand, which slowed troubleshooting and obscured where an order had stalled. Search by PO number or another shared identifier to retrieve every linked document at once, newest to oldest, with color-coded functional acknowledgment statuses showing at a glance whether an order has reached invoice, is sitting in shipment notice, or is waiting on an acknowledgment. Switch between Lifecycle View and the existing Transaction View from the dashboard. Learn more about tracking EDI documents using the lifecycle view.
MCP management - Caller context in MCP tools
MCP tools can now see who called them and through which server, so one tool returns results scoped to each person instead of the same data to everyone. Previously a tool had no way to know who invoked it, so you either exposed the full result set to every user or built a separate tool per user — row-level filtering over a shared connection wasn't possible. Reference the caller with context.byUser.email, context.byUser.name, and context.byUser._id anywhere you reference other tool values, filter by context.byUser.groups or context.byUser.roles, and identify the credential and server a call arrived with through context.token.type and context.source. Tools that don't reference context behave exactly as before. Learn more about caller context in MCP tools.
MCP management - MCP server registry API
Every environment now has an MCP server registry: one URL (https://api.integrator.io/mcp/<MCP account alias>/registry/v0.1/servers, or the same path on your region's API host) that lists, for each caller, the MCP servers it can connect to, so agents, scripts, and tools that read MCP registries discover their servers with an API token, a personal access token, or a Celigo Identity sign-in. Previously, an admin copied each server's URL out of integrator.io and a builder pasted it into every client, repeating the process whenever a server was added or someone's access changed. Now what each caller sees follows the access you already manage, personal access tokens can connect to MCP servers directly, only account owners and administrators can create, edit, or delete MCP servers through the API, and registry, oidc, and portal are reserved as account aliases and server paths. Learn more about discover MCP servers with the MCP server registry and MCP server registry API reference.
Celigo Storage activity log
Celigo Storage now records every upload, download, rename, move, copy, and delete with who did it, where it came from, and when, whether the operation came from the Storage UI, the API, a flow, an AI agent, the CLI, or a trading partner on your SFTP file server. Previously, Celigo tracked changes to your storage settings but not to your files, so "who deleted this?" and "did my partner send it?" meant opening a support ticket; now you can filter the log by source, actor, operation, outcome, or date range and export it, and partner file operations and every authentication attempt are attributed to the file server account that performed them. Activity follows your storage permissions, and entries start from the release with no backfill. Learn more about the Celigo Storage activity log.
SFTP file server permissions and SSH keys
Grant each Celigo-hosted file server account only the operations it needs, with independent Read, Create, Modify, and Delete permissions that support write-only drop boxes and read-only pickup folders. Previously every partner account had full access to its home folder and could authenticate only with a password; accounts now also accept SSH public keys (Ed25519, ECDSA, RSA 2048+), with multiple active keys for rotation without downtime, and users with Manage access on an integration can administer the accounts homed in that integration’s folders. Existing accounts keep their current full access until you change their permissions. Learn how to generate a key pair and manage SSH keys for a file server account.
Amazon Redshift as a Data Ingestion sync destination
You can now select Amazon Redshift as a managed Data Ingestion sync destination, alongside Snowflake, BigQuery, and NSAW. Celigo creates the destination schema and table on first sync, applies schema drift between runs, loads through Redshift's native COPY command with Celigo-managed staging, and upserts when a primary key is configured, with no AWS infrastructure to set up. See Create a sync to Amazon Redshift.
Connectivity - Shopify expiring offline access tokens
Celigo’s Shopify apps now issue and manage expiring offline access tokens rather than permanent ones, as Shopify requires of all public apps by January 1, 2027. Access tokens expire after 60 minutes and refresh automatically in the background, and the migration applies per store with no reinstall or merchant action needed. Without this change, Shopify would reject non-expiring tokens after that date and authentication would fail across every Shopify-connected flow. One behavior to know: a refresh token left unused for roughly 90 days expires, and the merchant has to re-authorize the connection.
MCP management - Upload MCP servers and richer Marketplace templates
You can now install an MCP server directly from a zip you downloaded with Download MCP, using the new Upload MCP action on the MCP Servers page. MCP server templates now carry a server's resources, capability sets, upstream MCP connections with proxied tools and prompts, and end-user credential requirements, so a published server installs the way it was built. End users, groups, roles, grants, and credentials never travel, and nobody has access to an installed server until you grant it; templates published earlier need a re-export to pick up the new contents. Learn how to upload an MCP server and install an MCP server template from the Marketplace.
AI automation - Network policies for webhooks and APIs
Network policies extend to webhook listeners and to API Builder and JavaScript API endpoints, so you can restrict which source systems trigger flows by webhook and which networks invoke your APIs. Blocked calls receive HTTP 403 and are recorded in request logs. Existing webhooks and APIs default to unrestricted, with no behavior change until a policy is assigned. See Create webhook listeners.
Connectivity - Flexible handlebars templates for JSON and XML file generation
File-based imports no longer require a specific template structure when generating JSON or XML. Write handlebars templates using any combination of helpers and expressions: {{#each batch_of_records}} and explicit header and footer structure are no longer required, though a non-blocking warning appears when each block is absent. Preview now validates JSON well-formedness. Learn about generating JSON and XML import files with handlebars.
Execution logs for tools in flows and APIs
A tool running inside a flow or API is now traceable from the flow run console or API run logs, including nested tools, following the parent’s log level, permissions, and retention. Previously a tool running in a flow gave you no visibility into how a record moved through it.
Integration apps - Ora-guided template discovery and install in the Marketplace
Templates that require Celigo Ora are now visually distinguished in the Marketplace and filterable, with a dedicated call to action replacing the standard install path. Previously they looked like any other template and offered an install flow that didn’t apply to them, which made them effectively unusable. Starting an install now hands off directly into Ora, pre-loaded with the selected template, and Ora reuses connections already in your account, prompting only for what’s missing. Discovery and install for standard templates and integration apps are unchanged.
API management - Immediate failure when a connection is offline
A new Immediate API run failure when connection is offline setting returns HTTP 500 as soon as any connection on the API is offline, rather than leaving the caller waiting up to two minutes for a timeout. It’s on by default for new APIs and off for existing ones, so enable it on APIs you want to fail fast. Note that the check covers every connection on the API, including steps the current request wouldn’t run.
Static lookup values in Handlebars preview
Previewing an HTTP import request body that uses a static lookup in a Handlebars expression now shows the value the lookup returns, matching what the flow produces at run time. Previously, Preview showed a placeholder such as {ASYNCID01} whether the expression was right or wrong, so the only way to check it was to save the resource and run the flow. This is a Preview-only change; run-time behavior is unchanged.
October 23, 2026: Integration apps and templates
Shopify–Microsoft Business Central template
A new pre-built template connects Shopify with Microsoft Business Central, covering both direct-to-consumer and B2B commerce out of the box, so merchants running Business Central as their ERP don't build the integration from scratch. See Shopify – Microsoft Business Central D2C B2B implementation guide.
October 22, 2026: Integration apps and templates
Support for Shopify expiring offline access tokens
Shopify connections in Celigo integration apps, such as the Shopify - NetSuite integration app, now use Shopify’s expiring offline access tokens, which Shopify requires for all public apps by January 1, 2027. Previously, Shopify connections used access tokens that never expired; now tokens expire every 60 minutes, and integrator.io renews them automatically in the background. Celigo migrates eligible connections between November 1 and November 15, 2026, before the Black Friday/Cyber Monday period, with no downtime or reinstall. Because Shopify allows only one active token per store, each store can be connected from only one integrator.io account. A store still connected from more than one account on November 1, 2026, isn’t migrated, and its connection stops working on January 1, 2027, so those customers work with their Celigo account manager to consolidate to one primary account. See FAQs: Shopify expiring offline access token migration.
Inventory details and bins carry forward from transfer orders in the Shipments (TO) add-on
In the Amazon - NetSuite integration app, the Transfer Order (TO) - Shipment add-on can now read the inventory details already set on your NetSuite transfer orders, including lot numbers, serial numbers, expiration dates, and bin assignments, and carry them through to the Item Fulfillments and Item Receipts it creates. Previously, the add-on assigned lots and serials from available stock at the location, ignored the transfer order’s inventory details, and didn’t support bin-tracked items. That meant sellers shipping lot-tracked, serialized, or bin-managed inventory to FBA had to create these records manually. Now, when Amazon splits a transfer order into multiple shipments, lots are allocated first-in, first-out (FIFO) and serials in order, with nothing assigned twice. New Default fulfillment bin and Default receiving bin settings handle bin-tracked items, and missing inventory details produce errors that name the transfer order, the item, and the setting to fix. Existing integration apps work as before because Auto-assign number to Lot Numbered/Serialized items stays selected; clear it on the General tab of Settings > Shipments (TO) to carry inventory details forward. New installations carry inventory details forward by default. See Manage your FBA shipments and NetSuite transfer orders (add-on).
October 15, 2026: Shopify OAuth apps, Model lifecycle status, MCP updates
Customer-owned Shopify OAuthapps for self-service connections
Self-service Shopify OAuth connections now use your own private Shopify app and credentials rather than Celigo’s shared public app. Shopify’s expiring token model supports only one refreshable token chain per app and store combination, so a shared app would put every customer on the same chain; a customer-owned app isolates yours. Existing self-service connections need to be re-authorized against a local app — the public app route stays available as a fallback until mid-December 2026. Integration app connections are unaffected and remain on the Celigo public app. See Set up an OAuth 2.0 connection to Shopify.
Model lifecycle status in the agent model picker
The model picker for AI Agents and Guardrails now shows each model's lifecycle status (active, deprecated, sunset-scheduled, or disabled). Deprecated or sunsetting models show a message when you configure, save, or run, with no silent fallback, and retired models are blocked with an actionable message. Switch any agent on a deprecated model to a current model before its sunset date. See Understand AI model lifecycle labels
OAuth 2.0 for Google Cloud Storage connections
You can now connect to Google Cloud Storage by signing in with Google. Previously, a Google Cloud Storage connection required a downloaded service account key, which some organizations don't allow. Now you choose OAuth 2.0 or Service account on the connection; with OAuth 2.0, enter a Default project ID, sign in with Google, and approve access. Existing service account connections keep working, and you can switch a connection between the two types and test it again. See Set up a connection to Google Cloud Storage.
Optimized bulk load for Amazon Redshift imports
Redshift imports now load through the native COPY command instead of row-by-row and batched SQL inserts, so loads that took hours finish in minutes. A new Use optimized bulk load (recommended for larger imports) query type is the default for new imports and needs no AWS setup at all — no S3 bucket, no IAM role, no cluster changes — since Celigo stages data in its own encrypted, short-lived area and never touches your AWS account. Optional primary keys turn the load into an upsert, and the existing Use bulk insert SQL type is renamed Use batch insert SQL, unchanged. Existing imports keep their current query type. See Import data into Amazon Redshift.
Real-time change data capture for MySQL
Real-time CDC Exports — Stream inserts, updates, and deletes from MySQL the moment they happen. Previously a MySQL export ran on a schedule and you tracked changes yourself with a timestamp or high-water-mark column; now change data capture reads the binary log directly, so MySQL joins MongoDB, SQL Server, and PostgreSQL as a real-time source. Select Listen for real-time data from source application as the export step type, scope the stream to specific databases, tables, and columns, and choose whether to snapshot existing rows first. Each event carries the operation type with the before and after versions of the row, and a stream resumes from its last processed position after a restart. Existing exports are unchanged; real-time exports require binary logging on the MySQL server and additional grants for the connection's database user. Learn how to listen for real-time changes to MySQL.
MCP server logging and retry job visibility
Open Run logs on a tool call in an MCP server’s request log to see its execution trace — connections used, mappings applied, upstream request and response, and the step that failed — where the log previously showed only that a call had errored. MCP servers add a per-server log level and a Capability column for filtering by the tool, prompt, or resource each request invoked. Delete log is replaced by Purge input/output and detailed log data, which strips payloads while keeping the time, method name, status, and caller, so sensitive data comes out without losing audit history. Retry jobs in run history now capture execution logs the same way standard runs do, where previously their logs showed no data and you couldn’t confirm whether a record was resent. Learn how to view MCP server request history and execution logs.
MCP end-user roles, capability sets, and group sync
Bundle MCP server and capability set assignments into a reusable role and assign it to a group or individual in one action. Capability sets now cover prompts and resources alongside tools and APIs, and are manageable from account management. Group sync resolves end-user group membership from Microsoft Entra ID, Okta, OneLogin, JumpCloud, Duo, and Auth0; Google Workspace can't send group membership over OIDC, so invite those end users directly. You can also reset an end user's MFA, and deleting a capability set converts access to direct grants rather than stripping it. Group sync matches against Celigo groups you've already created, so create them first. Learn how to create and manage end user roles.
Automatic page splitting for lookups over 5 MB
Previously, when a lookup pushed a page past the 5 MB limit, the records past that point errored and you had to retry them in small batches or shrink the page size and rerun the flow. Now integrator.io splits the page between records and keeps going. A single record that's over 5 MB with its lookup data still errors.
October 12, 2026: Integration apps and templates
Shopify–NetSuite billing flow GraphQL migration
The billing sync flow from NetSuite to Shopify moves from the REST transactions endpoint to the Shopify Admin GraphQL API, keeping billing flows working as Shopify deprecates its REST Admin API. Learn more about configuring the GraphQL billing flow.
Shopify–NetSuite SuiteTax compatibility
The Shopify–NetSuite integration app now supports NetSuite SuiteTax, so merchants using SuiteTax as their tax engine can sync orders, refunds, billing, returns, and cancellations without tax handling failures. Previously, the shared eTail engine never enabled SuiteTax behavior for Shopify, which meant import failures, incorrect tax postings, and manual workarounds for anyone on a SuiteTax account. SuiteTax handling now applies across order import, auto-billing for cash sales and invoices, refund import, returns and RMA, cancellation, and order export from NetSuite to Shopify. All three NetSuite tax preferences are supported — NETSUITE_CALCULATES, OVERRIDE_TAX_TOTALS, and TOTAL_TAX_AGAINST_SINGLE_LINE — including tax-inclusive pricing and the breakUpShopifyTaxLines setting. Learn more about SuiteTax support and updating the order export saved search for SuiteTax.
Variant-level sales channel publishing for Shopify–NetSuite
Merchants control sales channel publishing at the product variant level rather than only at the product level. Learn more about variant-level publishing.
Shopify–NetSuite refund import fix for third-party return apps
Resolves under-credited NetSuite credit memos for multi-unit refunds processed through third-party return apps such as ReturnZap and Loop Returns. Learn more about refund processing behavior.
Shopify–NetSuite customer deposit fix
Resolves a long-standing issue where customer deposits were never created in NetSuite for Shopify orders imported while payment was still authorized but uncaptured. Learn more about customer deposit creation.
Shopify–NetSuite GL account dropdown fix for GraphQL payout mappings
Resolves an issue where the NetSuite GL account dropdown in the Business entity transaction type to account mapping setting showed only the first 100 accounts. Merchants couldn’t select the accounts they needed when configuring GraphQL payout mappings. The dropdown now includes all eligible accounts, and the refresh controls for Business entity and NetSuite GL account have moved from individual mapping rows to the column headings. Learn more about the refresh options.
Shopify–NetSuite GraphQL payout account mapping fix
Resolves issues with the Shopify payout to NetSuite payout and deposit flow when account mappings are missing. The Business entity transaction type to account mapping setting isn’t populated automatically and must be configured manually. Before enabling the flow, ensure that you must map Charge, Refund, Adjustment, and Variance (catch-all) to NetSuite GL accounts for every business entity that processes payouts, including the primary business entity. Learn more about configuring payout account mappings.
Magento 2 NetSuite integration app SuiteApp migration
The Magento 2 NetSuite integration app moves from bundle-based distribution to a NetSuite SuiteApp, meeting NetSuite's current distribution requirement. Celigo migrates existing customers, and new customers onboard through the SuiteApp. Learn more about Magento 2 - NetSuite integration app SuiteApp migration.
Salesforce NetSuite integration app SuiteApp migration
The Salesforce NetSuite integration app moves from bundle-based distribution to a NetSuite SuiteApp, meeting NetSuite's current distribution requirement. Celigo migrates existing customers, and new customers onboard through the SuiteApp. Learn more about Salesforce - NetSuite integration app SuiteApp migration.
October 5, 2026: BYOK agents run code, alerts for long running flows, Mapper & Transformations improvements
Note: These features will roll out throughout the week of October 5, 2026.
💻Code execution for BYOK AI agents
Agents on a bring-your-own-key model connection can run code as a built-in tool, so they compute, transform data, or run logic mid-task without a custom script step.
⏰Alerts for long running flows
Configure email alerts for flows that run longer than expected, rather than discovering a stuck or slow flow from a downstream partner. Set a runtime threshold per flow or as an integration-wide default, where the per-flow value takes precedence. A historical runtime chart covering the last 24 hours, 7 days, 30 days, or 90 days helps you pick a defensible threshold based on how the flow has actually behaved. Learn more about setting alerts for long running flows.
↔️Mapper 2.0 and Transformations 2.0 improvements
Each mapping row gains an overflow menu with Clone row, Clone object/objectArray, and Set inactive. Cloning copies the configuration and leaves destination fields empty, while inactive rows and their children are skipped in preview and flow execution and marked visually; required rows can't be cloned or deactivated. You can name each object inside an object array rather than telling them apart by $, and fields added under one nested object stay scoped to it instead of applying to its siblings. A new Cause record to fail option appears under Action to take if source field has no value in Mapper 2.0 and Action to take if input field has no value in Transformations 2.0, for all import adaptor types. Preview now reflects your latest mapping and body parameter edits immediately, and empty mapping rows are dropped on save. Existing mappings keep their current behavior. Learn more about Mapper 2.0 and Transformation 2.0.
🎫Service principal authentication for Azure Synapse
The Azure Synapse connector now supports Microsoft Entra service principal authentication using the client secret method. Previously, teams whose Entra ID policies prohibit database-level SQL authentication had to fall back on the Universal JDBC connector with an on-premise agent; now they connect natively. Azure Synapse connections add an Auth type field offering Database user (existing behavior) or Service principal, and service principal connections reuse the same iClient configuration already available on the MS SQL connector. Existing connections continue to work as database user by default, so no migration is required. Learn more about creating an Azure Service Principal iClient.
📜Certificate-based service principal authentication for MS SQL
The MS SQL connector supports certificate-based authentication for Microsoft Entra service principals, matching Microsoft's guidance to use certificates rather than client secrets. Learn more about the MS SQL connection.
🎚️Simple and HTTP views for webhook listeners
Webhook listener forms add a Simple/HTTP toggle matching the pattern on HTTP connections and exports. Named providers such as Slack, Shopify, GitHub, and Stripe — more than 30 in all — default to the Simple view showing only the required fields, while generic webhooks keep the HTTP view. Switch to HTTP view at any time for verification, challenge responses, and other HTTP-level settings; toggling reveals all stored configuration values. Existing listeners continue to work without changes. Learn more about universal vs. application connectors.
🏷️XML parser metadata in the HTTP connection view
Exports built on an HTTP connection can configure a custom XML parser directly on the export form. Previously, exports returning XML failed with parser errors because the HTTP connection view had no way to attach parser metadata, forcing a switch to the Simple connection type and back. Learn more about creating an HTTP connection.
🔏PGP encryption for transferred files
PGP-encrypt existing files on transfer, not only files integrator.io generates from records. Imports with Generate files from records set to No now honor Encrypt files, across supported file providers rather than FTP alone, using the connection's existing PGP settings. If you've been pre-encrypting externally, validate output at the destination before removing that step. Learn more about PGP encryption.
🪄Automatic request parameter configuration in API Builder
Query parameters, path parameters, and headers configured on the API request node are added as transformation rules automatically, removing a repetitive mapping step that every API needed before downstream steps could use those values.
B2B - Automated VAN mailbox creation
Setting up a VAN connection is now self-serve. Previously the connection form told you to email Celigo to have a mailbox provisioned by hand; now you create the connection and Celigo provisions the ECGrid account automatically, then emails you its details and setup guidance so you can finish configuring the mailbox yourself. Saving the connection confirms what to expect next rather than leaving you waiting on support. Learn how to set up a VAN connection.
🕵️Rate limit detection for GraphQL connections
HTTP connections can detect throttling that a GraphQL API reports inside the response body rather than through the HTTP status code. Some GraphQL APIs, most commonly Shopify's, return HTTP 200 even when a request has been throttled.
🖥️Fit-to-screen error management grids
Error management grids auto-fit column widths to your screen by default, so a wide Message column no longer forces horizontal scrolling on smaller displays.
🐞Step-level debug logs in Tool Builder
Turn on and view runtime debug logs for individual steps while building a tool.
September 25, 2026: Celigo storage query, prebuilt MCP servers in Marketplace
🧮 Query Celigo Storage files
Introduces a read-only REST API for querying and transforming data in Celigo Storage files with SQL. Run queries directly against CSV, JSON, NDJSON, and Parquet files stored in Celigo Storage without impacting existing flows. It eliminates complex multi-flow or data warehouse workarounds by supporting joins, aggregations, window functions, CTEs, subqueries, and folder references (e.g., /staging/orders/). The API delivers fully typed JSON outputs without silent row truncation, provides granular file-parsing configuration, and enforces pre-execution token scope validation to safely reject unauthorized or invalid queries. Learn more about querying Celigo Storage files.
🏪Prebuilt MCP servers in the Marketplace
Browse and search the Marketplace for prebuilt MCP servers, then open a listing to see its tools, prompts, and connections and read any individual tool's input and output schema before you commit. Previously, getting a useful MCP server in front of an AI assistant meant building it yourself, with no way to judge someone else's without recreating it first. Install into your account with prompts for the connection and configuration values the server needs, validated before it goes live, or publish your own server as a Marketplace template. Credentials never travel with a template — sensitive values become placeholders you supply at install. Servers you already built are unaffected. Templates cover tools and prompts; MCP resources aren't included yet, so a server that defines them publishes without them. Learn how to find and install an MCP server template from the Marketplace.
🌊Data ingestion - Chunk-based streaming for merge imports
Merge operations now load to your warehouse in 250 MB chunks, so records become queryable in the destination as each chunk lands rather than only when the full sync finishes. Previously a large merge sync meant hours with nothing available and a restart from scratch if it failed late in the run. Append and replace mode imports are unaffected.
🗂️API group assignment during create, clone, and move
Assign an API to a group when you create, clone, or move it, even before the relative URI is defined. Previously, new APIs landed in the Unassigned group and had to be moved afterward.
September 18, 2026: MCP and Data ingestion
🧩 MCP server upstream connections
Connect a Celigo MCP server to third-party native MCP servers such as Salesforce, HubSpot, and Snowflake, and govern all of their capabilities from one place. Attach an upstream MCP connection to bring in its tools, resources, and prompts, which are discovered automatically and can be enabled or disabled individually. Override the name, description, and annotations on a proxied capability so AI agents see your organization's guidance rather than the generic upstream text, and add Celigo-built capabilities alongside proxied ones on the same server. Your existing governance applies to every call: authentication, end-user groups, capability sets, and request logging. Existing MCP servers are unaffected. Learn more about the MCP server proxy.
🙋 MCP server end-user connections
End users authenticate to downstream apps with their own credentials. A connection can be marked as requiring end-user credentials, and child connections store only the end user's credentials while referencing the parent at runtime. Learn more about MCP server end user connections.
📥Data ingestion - NetSuite JDBC as a Celigo Sync source
Connect NetSuite JDBC as a native source application in the Celigo Sync wizard and select multiple NetSuite objects for a pipeline in a single pass, rather than hand-writing a separate query per object. For detailed information on sync, see Create a sync from NetSuite JDBC.
📥Data ingestion - Microsoft SQL Server as a Celigo Sync destination
Load data through Celigo Sync into Microsoft SQL Server, which joins Snowflake and NetSuite Analytics Warehouse as a destination for sync pipelines. For detailed information on the sync, see Create a sync to Microsoft SQL.
📥Data ingestion - Object-level selection for manual runs and full resyncs
Choose which objects to include when you start a manual run or a full resync on a sync, instead of running every object each time. Resync a single object after adding a column at the source, or re-run only the objects that failed. The selection list shows each object's last-run status, and objects you don't select keep their data, sync state, and last-run status untouched. Run details reflect only what you selected, so record counts, durations, and errors match the actual run. Running with all objects selected behaves as it does today, and scheduled runs continue to process every object on the sync. Learn how to choose objects to include in a sync.
September 16, 2026: SFTP file server, Celigo Topics, SCIM, MCP rate limits, network policies, and more
📁 Celigo-hosted SFTP file server
Give every trading partner a secure SFTP endpoint hosted by Celigo, with no servers to run on either side. Enable a hosted file server per environment, then create an isolated account per partner with its own password and a Celigo Storage folder as its locked home directory. Partners connect with any standard SFTP client on port 22, and the files they upload become ordinary Celigo Storage files the moment they arrive — visible in the Storage UI, governed by storage usage, and consumable by flows through the Celigo APIs connector. Learn more about the Celigo-hosted file server.
📡 Celigo Topics
🪪 SCIM provisioning
Automate user lifecycle management from your identity provider. Provision and deprovision the people who build and operate integrations from a new SCIM Provisioning section in the account's Security and SSO settings, and manage end users from a matching section in each external IdP configuration. Learn more about provisioning users with SCIM.
🚦MCP server rate limits and usage quotas
Configure three optional traffic limits on each MCP server. A rate limit caps the total tool calls the server accepts in a time interval, protecting downstream systems from bursts. A per-user rate limit gives every end user their own allowance, so a single runaway AI agent can't consume the server's capacity and degrade service for everyone else. A usage quota caps cumulative tool calls over a longer period. When a limit is reached, requests receive a standard 429 response telling the client when to retry, and throttled requests appear in the MCP server logs with the consumer's identity. All three are optional — leave them blank for unlimited access. Learn more about MCP server rate limits and usage.
Intelligent document processing for AI agents
AI agents extract structured data from invoices, purchase orders, and Excel files into a schema you define, with field-level confidence signals. Agents detect document extraction intent and invoke processing automatically, with no setup beyond the existing agent schema. Learn how to extract data from documents with an AI agent.
🧱Network policies
Administrators can now define reusable IPv4/IPv6 allow and deny lists under Settings > Security and apply them across workspace logins, MCP servers, end-user authentication, and API tokens without re-entering ranges (PATs inherit workspace policies automatically). Deny lists take precedence, empty allow lists leave access unrestricted, and enforcement only begins once a CIDR is configured, ensuring existing endpoints experience no immediate behavior changes. Blocked logins and calls are recorded in audit and request logs, and full CRUD API support is available for programmatic management. Learn more about network policies.
📑B2B - EDI support for file provider connectors
Parse and generate EDI X12 and EDIFACT documents on file provider connectors, including Google Drive, Box, Dropbox, Azure Blob Storage, Google Cloud Storage, and Celigo Storage. EDI profiles were previously available only on FTP, S3, AS2, and VAN, so EDI flows had to route through one of those systems regardless of where the trading partner actually exchanged files. Choose EDI X12 or EDIFACT as the file type on a file provider export, import, or lookup, then select an EDI profile exactly as you would on an FTP or S3 flow step. Parsing and generator definitions, processing of parsed data, and functional acknowledgement generation behave the same as on FTP and S3, and Celigo Storage now handles EDI, so you can stage EDI documents between flows without external file infrastructure. Configure it through the UI or the v1 API. Existing EDI flows on FTP, S3, AS2, and VAN are unchanged. Read about the EDI connectors for B2B manager.
🔐Tenant-wide SSO enforcement
A new account-level setting requires every user to sign in through the identity provider, so no one is left on optional SSO. Account owners can now require SSO for themselves as well, which was previously impossible and removes the last username and password path into the account; that change is confirmed with a live SSO re-authentication first to protect against lockout. If the identity provider becomes unavailable, access is restored through a support-assisted, identity-verified recovery process. Both settings are off by default. Learn more about requiring SSO for the account owner and all account users.
☑️ Object-level selection for manual runs and full resyncs
Choose which objects to include when you start a manual run or a full resync on a sync, instead of running every object each time. Resync a single object after adding a column at the source, or re-run only the objects that failed. The selection list shows each object's last-run status, and objects you don't select keep their data, sync state, and last-run status untouched. Run details reflect only what you selected, so record counts, durations, and errors match the actual run. Running with all objects selected behaves as it does today, and scheduled runs continue to process every object on the sync. Learn how to choose objects to include in a sync.
💾 Shift JIS file encoding on FTP and file connectors
Choose Shift JIS as the file encoding when you export, look up, import, or transfer files on FTP, SFTP, FTPS, and other file connectors. Japanese files encoded in Shift JIS show the correct characters in preview and in your flow, and files you write back use the encoding the destination system expects. See File transfers.
September 2, 2026: Celigo Storage, AI agent models
📁 Celigo Storage integration folders and token path scoping
Each integration now gets its own system folder, with permissions inherited from your existing access control, so Manage All, Monitor All, and Custom-role users reach Storage through their integrations rather than needing account-wide access. API tokens can be scoped to specific folder paths, limiting what a given credential can read or write.
Work with files and folders in integrator.io
Restrict an API token to specific Celigo Storage paths
Four more model providers for AI agents and guardrails
Why it matters: Model choice used to mean OpenAI, Gemini, or Anthropic. These four run on our new provider catalog framework (2026.8.2), which turns providers into catalog data instead of platform code — that's why all four landed at once with no release, and it means we can add new providers quickly as they're requested.
Who it's for: Builders who want the right model per agent, and IT admins standardizing on a provider their org already governs
What changed:
- Mistral, xAI, Hugging Face, and DeepSeek join OpenAI, Gemini, and Anthropic in the model picker — BYOK via your own provider connection, and AI Credits are not charged for BYOK runs
- Agent and guardrail settings adapt to the selected model — you only see the controls that model supports
- Existing agents are untouched, and invalid provider/model/option combos are rejected on save with an error that names the problem
August 31, 2026: NetSuite REST API support, Anthropic in Celigo AI
🛠️ NetSuite REST API Support (SuiteTalk + Custom RESTlets)
The NetSuite connector now supports REST APIs as a first-class API type for record-based exports, lookups, and imports — covering both SuiteTalk REST Web Services and Custom RESTlets, directly within the NetSuite connector without requiring a separate generic HTTP connection.
Why it matters:
NetSuite’s Saved Search — the basis of most existing NetSuite exports — is known to be inefficient at high volumes. SuiteTalk REST Web Services provide a faster, simpler alternative for large-volume and data warehousing workflows, and don’t require the additional SuiteAnalytics Connect license that the JDBC approach requires. REST also gives builders simpler access to custom records and fields. With this addition, Celigo now supports all three major ways to integrate with NetSuite: Celigo RESTlets, SOAP Web Services, and SuiteTalk REST.
Who it's for:
Builders running high-volume NetSuite integrations or data warehousing workflows, and developers who prefer REST over SOAP or need direct access to custom NetSuite record types.
What changed:
New API type in the NetSuite connector. A REST APIs option is now available in the “How would you like the records exported/imported?” selector on record-based NetSuite export, lookup, and import steps. Selecting REST APIs exposes two sub-options: SuiteTalk (configurable via a NetSuite-oriented Simple view or an HTTP view for advanced control) and Custom RESTlets (configurable via HTTP view). The platform automatically sets the appropriate NetSuite REST base URI based on the selection. File-based flows are unaffected. Existing Celigo RESTlets, SOAP Web Services, and generic HTTP-based integrations continue to work unchanged.
🛠️ Anthropic (Claude) in Celigo AI [Improved]
You can now run AI agents on Claude models without bringing your own Anthropic API key.
Why it matters:
Until now, using Claude in Celigo meant BYOK — your own Anthropic account, key management, and separate billing. That's gone: Claude is now hosted under Celigo AI, selectable just like OpenAI and Gemini, with usage billed through Celigo.
Who it's for:
Anyone building with AI agents, guardrails, or AI chat — especially teams without their own Anthropic agreement
What changed:
- Claude models (Opus, Sonnet, and Haiku) are now selectable under the Celigo AI source for AI agents, guardrails, and chat.
- Full parity with OpenAI and Gemini — tool calls (IO tools and MCP servers), structured outputs, streaming, extended thinking, and Anthropic web search.
- BYOK is unchanged — existing Anthropic, OpenAI, and Gemini setups keep working as-is.