What's New in Willow
Every capability, improvement and fix shipped to the platform. Governed access for AI agents, updated weekly.
One email per release. Unsubscribe anytime.
Offline-Ready Local Library
@mcp-s/mcp caches tools on the device, starts local servers on first use, and keeps working offline.
Read the docsStructured AI Risk Scoring
Consistent risk scores for shadow MCP servers and skills, alongside a unified Shadow AI agents view. Beta.
Read the docsGateway Bypass Detection
Shows whether discovered MCP calls went through the Willow gateway, bypassed it, or are unknown.
Read the docsBackground Agent Catalog
Agents are grouped by org groups, shown on richer cards, with a catalog for discovering available agents.
Read the docsAPI Trigger User Identity
Agents started through the API can act as a specific user via the x-user-email or x-user-id headers.
Read the docsProxy Bypass Wildcards
Outbound proxy bypass host lists now accept wildcard patterns, so whole internal domains skip the proxy.
Read the docsJumpCloud Device Attribution
Map devices to users automatically through a JumpCloud service account in AI Discovery settings.
Read the docsConfig Deployments
Push Claude Code, Claude Desktop, and MCP config files to devices by group and OS, with version history.
Read the docsPersonal Account Detection
Flags Claude Code and Cursor signed in with personal, non-company accounts on employee devices.
Read the docsOrganization Logo Upload
Upload your company logo with a 3:1 frame editor to brand the admin app and Connect portal.
Read the docsQuick MCP Setup
Set up an MCP server on one screen, with a connector tester, default OAuth scopes, and a custom server editor.
Read the docsClaude Login via Okta
Enterprise-managed authentication connects Claude to Willow through Okta Cross App Access, with no separate sign-in.
Read the docsRadar Attack Paths
Radar risks now show severity levels, grouping, and flow graphs explaining how each risk could be exploited.
Read the docsAgent Telemetry Settings
Admins configure the OTLP exporter variables that AI agents use to send telemetry to Willow.
Read the docsEmbedded AI Discovery
Discover and control AI assistants built into SaaS apps, like Linear AI, in Shadow AI and Prompt Guard.
Read the docsDevice Access Map
A 2D/3D map of each device's AI agents and the MCPs, skills, and plugins they can reach.
Read the docsTool Call Flow Graph
Each logged tool call drawn as a graph covering SSO, approvals, guards, credentials, rate limits, and the failure point.
Read the docsUnified Connectors Catalog
One shared catalog of API and official MCP connectors, now powering Add MCP Server for every org.
Read the docsCursor Post-Tool Guard
The guard hooks plugin now checks Cursor tool outputs after each call, not only the inputs.
Read the docsAgent Settings File Discovery
Collects each user's AI agent settings files, including inference provider config, shown as a JSON tree.
Read the docsYour agents are already in the wild.
Give them a Basecamp. Go from AI chaos to AI work, in minutes.
