October brings Claude Managed Agents that run end to end from Port workflows, Identity Propagation so integration actions run as the user who triggered them, and saved filter presets for dashboards. Guided Solutions Onboarding, 26 new workflow integration actions, and a wave of dashboard, connector, and governance improvements round out the month. Here’s everything that’s new in Port.
Big
Run Claude Managed Agents end to end from Port workflows
Manage the full Claude agent lifecycle, use the agent's results, watch it work live, and give it Port MCP access, all from Port workflows.
Port workflows can now create, update, and archive Claude Managed Agents and their environments, wait for a session to finish and use what the agent produced, and stream the agent's progress into the run log while it works.
Agents triggered from a workflow can also get Port MCP access automatically, acting as the user who triggered the run or as a service account, with no credentials to set up and store in the Claude vault.
image
Run integration actions as the triggering user with Identity Propagation
Self-hosted Azure DevOps, GitHub, and GitLab integration actions now run with the credentials of the user who triggered the workflow.
Get per-user attribution and least-privilege access: the third-party tool's audit logs show the real user, and each action is limited to what that user may do. User tokens stay in a vault you own. Set
OCEAN__IDENTITY_PROPAGATION__ENABLED=true
(plus vault and OAuth settings) on your self-hosted integration and
"enableIdentityPropagation": true
on the action node. Users sign in to the tool once, on their first run.
image
Save dashboard filters as presets
Admins can now build several filtered views on one dashboard, so you don’t need a separate copy for every variation.
Set up one dashboard and save a preset for each view you need, instead of duplicating it in the software catalog. For example, save presets for managers and for a specific team, or for last week, last month, and last quarter. Apply the filters you want, click
Save as preset
, then click
Save this view
to publish it to all. Viewers switch between presets in one click.
image
Medium
Track work in the Board widget
The new
Board
widget (aka Kanban) shows catalog entities as cards in columns, grouped by a property like status or priority.
See how work moves across stages for work items, services, or incidents handled by people and agents. Drag a card to another column to update its value. You can also switch an existing table to board view.
Availability:
Closed beta
image
Filter dashboards with advanced rules and one-click edits
Dashboard filters now support advanced rules, contextual queries, and one-click value editing.
  • Advanced mode:
    Combine multiple rules with And/Or, edit JSON for nested queries, or filter by the viewer’s User or Team properties (contextual queries).
  • Quick edit:
    Click a single-rule chip to pick a new value, or use the
    ⋯
    menu to duplicate, edit, or delete. Members can now adjust their view without opening the filter editor.
  • Placeholders:
    Save a
    has any of
    ,
    is in
    , or boolean
    =
    filter with no values, and members fill them in, the same way initial filters work on catalog pages and in widgets.
A dashboard filter in advanced mode:
image
💪 Built from your feedback:
Get started faster with Solutions Onboarding
New to Port? Pick the problem you want to solve, and Port sets up a working workflow and dashboard for it.
Instead of working out what to connect, model, and build first, new orgs follow a short guided path. Choose a use case (Autonomous ticket resolution, Self-healing incidents, Engineering intelligence, Cloud resource management, or AgentOps) and connect your Git provider and the tools that matter for it. Port provisions the data model while AI builds a runnable workflow and dashboard from the matching guide. A setup checklist then shows which integrations, secrets, and workflows are left, each one click away. Sign up or create a new org and follow the onboarding; when it finishes, open the setup portal button in the sidebar to see your checklist.
image
Automate more with 26 new workflow integration actions
Port Workflows adds 26 integration actions across GitHub, Jira, Azure DevOps, GitLab, Linear, PagerDuty, Claude, and Cursor.
Automate incident response and CI fix loops straight from the Workflows builder, without custom webhooks. For example, acknowledge a PagerDuty incident, open a Linear war-room issue, dispatch a GitHub rollback workflow, and trigger a Cursor or Claude agent to fix a failing GitLab pipeline. Pick the actions from the "Add node" modal under the relevant integration.
image
Respond to workflow input nodes from your AI agent
AI agents like Claude, Cursor, and Port AI can read and answer pending workflow input nodes through Port MCP.
Approve, reject, or fill in a workflow's input form from your agent without switching to the Port UI. Responses are now validated against the node's form everywhere, so missing required fields or wrong types return a clear error. Ask your agent about your workflow run, and it will read the pending input and submit your choice with the new
submit_input_node_response
MCP tool.
image
Catch credential and permission issues before you sync with Test Connection
Find credential and permission issues as soon as you install a connector, before any sync runs.
Until now, bad tokens, missing scopes, or the wrong org showed up only after a full resync failed, and your data went stale in the meantime. Test Connection checks a connector's connectivity and access for each kind without running a resync. Run it when you install an integration, before the first sync, and fix issues in advance. It's available today for AWS-V3, GitHub, Jira, and Plain.
image
Get Slack and email alerts when a connector sync breaks
Admins can now get connector sync alerts in Slack or by email, not only in the Port Inbox.
When a sync fails, your data goes stale, and admins aren't always in Port to see it. Now the admins or teams you choose get an alert in Slack or by email when a sync fails, a connector is suspended after repeated failed syncs, or a sync deletes more than 50% of the entities. Each alert links to the sync status so you can investigate. Slack and email alerts are opted out by default. To opt in, open the connector modal and turn them on under
Settings
→
Manage notification
.
image
Test mappings against 100 live examples
The Mapping tab in Edit Exporter now shows up to 100 live examples per kind, and you can filter them across everything your integration has ingested.
Write mappings that hold up against the payloads your integration actually ingests, edge cases included, instead of a handful of samples. The filter runs on all ingested data before examples come back. That means you can find the exact payloads you need to check even when they’re rare, like pull requests from one specific author. Open the
Connectors
→
Edit
→
Mapping
tab, then set a filter under
Mapping Examples
in the transformation form or the YAML playground.
image
Small
Limit AI agents to specific workflows
Choose exactly which workflows an AI agent can run, in the agent form or via the
allowed workflow triggers
property, through either the UI or API.
Keep each agent scoped to the tasks it was built for instead of giving it access to every workflow. Port blocks any other workflow, even if a user approved the run.
image
Control who can sign in to your Port organization
You can now make SSO the only way into your Port organization.
Once your domains are verified, turning on 
Require login using SSO
 in the SSO tab of organization settings forces everyone on those domains to sign in through your identity provider, blocks password sign-ups, and prevents anyone outside your verified domains from being invited or signing in.
Monitor runs across workflows with the Workflow runs history widget
A new dashboard widget shows runs from one, several, or all workflows in a single filterable table.
Keep a rolling "what needs attention" view instead of opening each workflow's run history one by one. Scripts and AI agents can fetch the same views through new API filters. Open a dashboard → Add widget → Workflow runs history → pick workflows (or none for all) → optionally save default status and date filters.
GET /v1/workflows/runs
now accepts
createdAfter
,
createdBefore
,
displayStatuses
, and
excludeNodeRuns
.
Turn a single widget into a full-page view
When a page holds a single widget, you can now hide its border and header for a clean, full-page view. This works with any widget, like a table or a custom widget.
It’s most useful for custom widgets. Build a plugin, connect it to a custom widget, and the page looks like a polished report or a standalone app. To turn it on, open the widget’s
⋯
menu and select
Expand to Page
.
image
Drill down into pie and bar charts by Scorecard
Pie and bar charts that use
Breakdown by property
with a scorecard now support drill down view. Click a slice or bar to see the entities behind it.
A chart shows how many services meet a standard, and now one click shows which services meet it and which don’t. Go straight from “12 services at Bronze” to the list of those services and act on them.
sc-drill-down-gold
Build and share workflows more easily
Deep-link to workflow runs from messages, preview markdown in forms, hide triggers from the Self-Service Hub, and use a cleaner node menu.
Send approvers straight to the right step, review long markdown text at a glance, and keep the Self-Service Hub focused on the triggers users should start from there. Add
{{ .workflowNodeRun.url }}
or
{{ .port.appUrl }}
to any node's message; markdown fields render as a preview automatically; turn off "Show in Self-Service Hub" in a trigger's Contexts section (or set
{ "on": "SELF_SERVICE_HUB", "visible": false }
).
💪 Built from your feedback:
UI inactivity timeout
You can now set a timeout that automatically signs users out of the Port UI after a period of inactivity.
Claude Sonnet 5.5 is now the default model in Port AI
Port AI now runs on Claude Sonnet 5.5, replacing Sonnet 4.6 as the default. Responses start about twice as fast, and multi-step agentic tasks finish in less than half the time, so your teams wait less on every answer and every workflow run.
The new model is available to all organizations using the Port provider. If you haven't manually picked a model in your AI settings, you're already on Sonnet 5.5 with nothing to do. If you chose a specific model, your setting stays as it is.
For the best performance and product experience, Port recommends switching to Claude Sonnet 5.5.
image
Connect Port AI to AWS Bedrock with an API key, or through your own gateway
Port AI now supports Amazon Bedrock API keys, so you can connect Bedrock without setting up IAM roles or access key pairs. Store the key as a Port secret and you're connected.
If your organization routes LLM traffic through an internal gateway or proxy, you can now point Port at it with a custom base URL. This works with any of the three Bedrock auth methods (API key, assume role, access keys). This helps platform teams get Port AI running on Bedrock faster, and keep model traffic inside the controls they already have.
💪 Built from your feedback:
Set custom properties and relations on port_scorecard in Terraform
The
port_scorecard
Terraform resource supports custom properties and relations on scorecards and each rule (released in provider v2.31.0).
Keep the whole scorecard definition, including custom metadata, in one Terraform resource. Extend the
_scorecard
and
_rule
system blueprints, pass
properties
and
relations
as
jsonencode({...})
, use
depends_on
, and upgrade to provider v2.31.0.
Show all entities in entity page widgets
Table, pie chart, and number chart widgets on an entity page can now show entities beyond the ones related to that entity.
Turn off
Show only entities related to…
in the widget settings form to show all entities of the selected blueprint. Use it to add broader context next to an entity, like all open incidents across the org on a service page. Leave the toggle on to keep the widget scoped to that entity’s relations.
image
Choose which tabs appear in global search
Admins can now choose which tabs appear in global search.
Go to
Settings
→
Search
and turn tabs on or off under
Global search tabs
. The options are
All
,
Entities
,
Pages
,
Actions
,
Tech Docs
, and
Blueprints
. A hidden tab is hidden for all users, including admins. For example, hide
Blueprints
so the search shows only the results your users need.
image
Guides
New guides published this month: