You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This commit was created on GitHub.com and signed with GitHub’s verified signature.
Minor Changes
55a25ac: Add management APIs and dashboard UI for enabling and configuring MCP server tool filtering via tool variation groups.
Patch Changes
8f3591d: resolve /mcp/ OAuth flow handlers via mcp_endpoints with toolset fallback
a1f25dc: Prepare RBAC grants for issuer-gated private remote MCP servers so tools/list and tools/call no longer fail for RBAC-enforced callers. Previously the issuer-gated path skipped grant preparation, causing the proxy's mcp:connect interceptors to reject the request with a missing-grants error and return zero tools.
13551ec: Add the assistant_dashboard_messages table — the user-visible conversation log for the AI Insights sidebar (user messages + the assistant's delivered replies), kept separate from the raw chat_messages transcript. Keyed by chat with a monotonic seq for incremental polling. Foundation for AGE-2631.
3011492: Add an endpoint for a dashboard user to send a message to an assistant. The reply is delivered asynchronously — the response returns the chat to poll for it. The caller chooses the conversation thread via a correlation key (send the user id for one continuing thread per user, or a fresh value to start over), and can pass an idempotency key so a retried send doesn't enqueue the message twice.
1078e46: Add an optional user_id filter to the risk events list. The Risk Events page now exposes a "User contains..." search box that filters findings by the chat's external user id (case-insensitive substring match), alongside the existing policy and rule filters.
3eaa1cf: Add message_types to risk policies so admins can target enforcement and batch scanning to user messages, tool requests, tool responses, or assistant text.