| name | update-stacks |
| description | Use when the user asks to update, deploy, or apply changes to all Pulumi stacks (core and all stacks in the stacks/ folder). Trigger on phrases like "update all stacks", "deploy all stacks", "pulumi up all", or "apply all stacks". |
Update Stacks
Updates the core stack and all child stacks in the stacks/ folder.
Workflow
1. Build the project
Compile shared packages first:
npm run build
2. Discover stacks dynamically
- Core stack: Look for
Pulumi.<stack>.yaml in the project root. Use that stack name.
- Child stacks: For each subdirectory in
stacks/, look for Pulumi.<stack>.yaml. Collect all stack names. Each child stack must be run from its own directory (stacks/<name>/), not the project root.
3. Preview all stacks
Run previews in parallel where safe:
- Preview the core stack first (alone) — run from project root
- Preview all child stacks in parallel — each must be run from its
stacks/<name>/ directory
Core stack:
pulumi preview --json -s <stack-name>
Child stacks (run from stacks/<name>/):
pulumi preview --json -s <stack-name>
If a stack has no changes, note that. Otherwise, extract from JSON:
create, update, delete, replace counts- Any policy violations or errors
4. Summarize and ask for confirmation once
Present a table:
| Stack | Create | Update | Delete | Replace | Status |
|---|
| lab (core) | 2 | 5 | 0 | 1 | changes |
| lab-apps | 0 | 0 | 0 | 0 | no changes |
| lab-ai | 1 | 3 | 0 | 0 | changes |
Then ask: "Apply all stacks?" (yes/no). Do not proceed without explicit confirmation.
5. Update core stack first
Only if core has changes from preview. Run from project root:
pulumi up --yes -s <core-stack-name>
Wait for completion. If this fails, stop and report the error. Do not update child stacks.
6. Update child stacks in parallel
Only run on child stacks that have changes from preview. Run them simultaneously from their respective stacks/<name>/ directories:
pulumi up --yes -s <child-stack-name>
7. Report results
Present a final summary:
| Stack | Result | Notes |
|---|
| lab (core) | ✅ success | |
| lab-apps | ✅ success | no changes |
| lab-ai | ❌ failed | error message |
For failures, include the last few lines of the error output so the user can diagnose.
