MCP Tools, Smarter Investigations, and Workspace Defaults
The August 2026 weekly v2026.8.9 Aiden release exposes SRE and Platform capabilities over MCP, makes investigations reuse prior work instead of burning duplicates, and lets you pin the workspace that opens after sign-in. Remote runners can attach GitLab Vault secrets, Bedrock setup no longer pre-selects untested models, and invited workspace admins can reach settings again.
Explore the sections below to see what's new, enhanced, and fixed in this update.
| Area | Update | |
|---|---|---|
| What's New | MCP | SRE MCP Tools |
| MCP | Platform MCP Tools | |
| Workspaces | Default Workspace after Sign-In | |
| What's Enhanced | SRE | Smarter SRE Investigations |
| What's Fixed | Models | Adding an LLM Provider Pre-Selects Models |
| Remote Runners | Remote Runner Secrets Omit GitLab | |
| Workspaces | Workspace Context and Admin Settings |
What's New
SRE MCP Tools
Click to view
Aiden for SRE now exposes its core workflows as MCP tools, so editors and other MCP clients can work alerts, investigations, and discovery without leaving the IDE.
Tools include listing and syncing alerts, starting and reading investigations, and running discovery (list entities and insights). Names on the wire use the sre__ prefix.
See Configure MCP.
Platform MCP Tools
Click to view
Platform tools are available over MCP so clients can search Knowledge Hub runbooks, chat with an agent, resolve human-in-the-loop approvals, and answer clarification prompts.
Names on the wire use the aiden__ prefix (for example aiden__search_runbooks, aiden__chat_with_agent, aiden__resolve_approval, and aiden__respond_clarification).
See Configure MCP.
Default Workspace after Sign-In
Click to view
You can pin which workspace opens after you sign in when you belong to more than one.
Previously, login always landed on a default workspace you did not choose. Now, set the workspace that should open after sign-in and keep landing there until you change it.
See Workspaces.
What's Enhanced
Smarter SRE Investigations
Click to view
SRE investigations reuse prior work and stop closing as done before triage is really finished.
Previously, the same alert could spawn duplicate investigations, a run could mark complete without an accepted triage, and weak or mis-scoped monitors could burn tool budget without a clear early stop. Now, launch policy and dedup fingerprints cut duplicate reruns, sessions require a terminal triage before they complete, prior-session and correlation gates run when they help, and mis-scoped monitors can exit early on signal quality. Activity lists can show the Slack thread ID for a session.
See Investigations and Alerts.
What's Fixed
Adding an LLM Provider Pre-Selects Models
Click to view
Previously, adding Bedrock (or another LLM provider) pre-selected recommended models that were not verified against Aiden evals. Now, the form does not pre-select models. You choose the models to enable.
Remote Runner Secrets Omit GitLab
Click to view
Previously, the Remote Runners secrets UI only offered a short hardcoded list of integrations, so Vault GitLab secrets (and other subcategories) could not be attached on create or after create. Now, typed slots follow real Vault metadata, so GitLab and other integration secrets appear when they exist. You can attach them when you create a runner or from the secrets dialog afterward, including generic secrets.
Workspace Context and Admin Settings
Click to view
Previously, opening another Enterprise account in a new tab could drop the current workspace context, and an invited workspace admin might not see the bottom-left settings control. Now, workspace context stays with the tab, and invited admins can open settings.