What's new · November 2026
November 2026
Morph now asks for your help when it hits a blocker, repositories can now be added after starting a project, and much more
Morph now asks for your help when it hits a blocker, repositories can now be added after starting a project, and much more
Morph now asks for your help when it hits a blocker it can't solve alone
Some blockers might be caused by access only you have: a private package registry, an external database, fresh credentials, or your VPN. It now pauses and asks you for help. Morph then picks up right where it left off once you've sorted it out.

When Morph needs you, you'll see a "Morph needs help" banner in the app and get an email. Open the help chat to see what it needs. The chat works in the same workspace as the run, so it already has the full context and you can solve the problem together.

Once it's sorted, Morph picks up right where it left off. You decide how it continues:
- Fix it and continue: sort out the access, and the run carries on.
- Proceed without it: let Morph keep going without the missing piece.
- Lower the validation level: skip the checks that need what's missing, without restarting.
- Move to self-hosted: switch to your own ModelDaemon inside your network. Morph saves the work done so far, and the next run continues from there.
Add repos during project planning
You can now add a repo to your project during planning. If you realize partway through that you left one out, add it and Morph updates the plan to include it. There's no need to start a new project.

Morph brings the new repo into the plan and updates the milestones that depend on it, so the plan reflects everything your modernization touches.
- Morph updates the plan to account for the new repo.
- Keep the planning work you have already done.
Catch self-hosted setup issues before a run starts
ModelDaemon now runs a deeper set of checks on your machine before Morph starts work. Network, registry and access problems now show up during setup, with a clear reason, instead of halfway through a run.
Basic connectivity checks can pass while real installs still fail behind a proxy, mirror or firewall. ModelDaemon now tests the same downloads and access a run will actually need, first as soon as it connects and again once the environment is set up.
- Tests real package downloads from npm, pip and conda registries, using both your configuration and the registry Morph sets up.
- Confirms the machine can read your repos with the credentials Morph uses.
- Installs the tooling Morph needs automatically instead of stopping.
- If a setup step fails, Morph shows which step failed and why in your prerequisite checks.
See how your ModelDaemon machine is doing
You can now see how busy each self-hosted ModelDaemon machine is, right in the project sidebar. If a run feels slow, you can quickly tell whether the machine running it is the bottleneck.
Open a daemon's details to see its CPU usage over the last 24 hours, with any interruptions marked on the graph.
- Current CPU usage for each daemon, plus the busiest machine in the pool at a glance.
- Interruptions from the last 24 hours, such as when the machine slept, froze, restarted or disconnected.
- For offline daemons, it shows the last reading before they went quiet.
- Available for daemons on version 2.0.32 or later.
Knowledge chat is faster, smoother and smarter
Knowledge chat now responds the moment you hit send and opens past conversations in under a second.
Open a conversation you had yesterday, and it's just there. Your history now loads in under a second, even while Morph reconnects in the background.
Ask a question and you'll see Morph get to work right away. Instead of a blank pause, each setup step shows up as it happens, and the reply starts streaming more than twice as fast as before.
Starting a brand-new chat is quicker too. Morph prepares your repos in the background while the conversation begins, so you're talking about your code sooner.
When the answer finishes, the chat is ready for your next message almost immediately. Along the way, we also made chat more reliable and better informed:
- Close the tab mid-reply and nothing is lost. Morph finishes the turn on its own.
- Long documents are written in parts, and if a reply stalls, chat tells you why instead of hanging.
- When a knowledge doc changes, every chat in the project hears about it.
- Chat can now look things up in the Morph product docs to answer questions about the product.
- Review chat answers get the same safety checks as Knowledge chat.
- You can now use Knowledge Chat to directly start milestone executions, help retry failed runs, and understand milestone progress
General improvements and quality-of-life updates
Quality-of-life updates across models, milestones and chat.
- Wiki and Rules stay in sync: when a spec or milestone change contradicts your Wiki or Rules, Knowledge chat updates them to match.
- Change the repo map through chat: Knowledge chat is now the place to change how your repositories map together.
- Fewer surprise re-plans: Morph no longer forces milestone re-planning.
- Queued milestones are locked, so they can’t change underneath a run. Morph also reminds you once if you have unsaved environment edits.
- Lifecycle scripts run as soon as you add or edit them, so you know right away whether they work.
- More resilient runs: if an agent session fails, Morph resumes it once on the same setup before falling back.
- Review chat history is kept for 30 days.