Help us improve
Share bugs, ideas, or general feedback.
From sapcc
Autoscaling operations via Castellum. Triggers: autoscaling, castellum, resize, scaling, threshold, auto-resize, capacity management. NOT for: manual quota changes (use sapcc-quota/Limes).
npx claudepluginhub notque/openstack-agent-toolkit --plugin sapccHow this skill is triggered — by the user, by Claude, or both
Slash command
/sapcc:sapcc-autoscalingThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Inspect Castellum autoscaling state: view resource configurations, check pending resize operations, diagnose failed scaling actions.
Guides technical evaluation of code review feedback: read fully, restate for understanding, verify against codebase, respond with reasoning or pushback before implementing.
Share bugs, ideas, or general feedback.
Inspect Castellum autoscaling state: view resource configurations, check pending resize operations, diagnose failed scaling actions.
Note: Castellum MCP tools are planned but not yet implemented in the MCP server. The skill documents the service for reference. When tools become available, they will follow the
castellum_prefix pattern.
| Tool | Purpose | Expected Parameters |
|---|---|---|
castellum_list_resources | List autoscaling-managed resources | project_id |
castellum_get_resource | Get autoscaling config for a resource | asset_type, asset_id |
castellum_list_operations | List scaling operations history | asset_type, state |
Until Castellum MCP tools are available:
maia_query with PromQL to check Castellum metrics: castellum_resource_*hermes_list_events(target_type=autoscaling/resource) for scaling audit trailproject_id must be a valid UUID. The tool validates UUID format and rejects anything else. You cannot pass a project name — resolve it to UUID first via Keystone if needed.
asset_type uses a colon-separated hierarchy. Format is project-quota:<service>:<resource>, e.g., project-quota:compute:cores or project-quota:network:floating_ips. Getting this format wrong returns empty results without an error.
Castellum watches Limes, does not replace it. Castellum monitors usage via Limes and triggers resize operations. The actual quota values live in Limes. To see current quota/usage, use limes_get_project_quota. Castellum shows the autoscaling configuration and pending/failed operations.
Pending does not mean stuck. Pending operations are queued resizes that have not yet been executed. Castellum processes these asynchronously. Only flag operations that have been pending for an unusually long time (hours).
Failed operations include the failure reason. The recently-failed endpoint shows why a resize failed (e.g., parent quota exhausted, constraint violation). This is the primary diagnostic tool for scaling failures.
Not all resources are auto-scalable. Only resources that have Castellum configuration will appear. If get_project_resources returns empty or missing resources, those resources are not configured for autoscaling.
Thresholds define when scaling triggers. Each configured resource has usage thresholds (high/low) that trigger upscale/downscale. The get_project_resources response shows these thresholds alongside current usage.
castellum_get_project_resources with project_id=<uuid> — see all configured resources and their thresholds.limes_get_project_quota to see actual quota values.castellum_list_recently_failed_operations with project_id=<uuid> — check for failures.castellum_get_project_resources — check if thresholds are configured and if usage is actually above the high threshold.castellum_list_pending_operations — see all queued resizes (optionally filter by project or asset_type).| Need | Service | Tool |
|---|---|---|
| Current quota and usage values | Limes | limes_get_project_quota |
| Project UUID from name | Keystone | keystone_list_projects |
| Audit trail of resize actions | Hermes | hermes_list_events(initiator_name=castellum) |
| Compute resource details | Nova | nova_list_servers |