Skip to main content
Glama

Retire gone machine rows

retire_gone_machine_rows
Destructive

Retire platform server records of an Ankra-provisioned cluster (Hetzner, UpCloud, DigitalOcean, AWS, or Ankra Cloud) whose machine the cloud provider no longer has - the fix for a phantom worker or a stale server record get_cluster_machine_inventory classified 'phantom' or 'gone_kept'. Pass the resource_ids it listed in retirable_resource_ids (at most 20 per call). The scheduler proves every row gone again at write time and retires it ONLY when the provider answers 404 for the row's recorded machine id AND answers another machine of the same cluster as present in the same pass; any other answer - a timeout, an auth error, a 404 with nothing to confirm it, or the machine still present - leaves the row untouched and says why. A live worker record becomes down and permanently deleted together with its per-node Kubernetes record, and a node group without autoscaling whose last worker it was releases its settings, as a normal worker delete does; a parked worker record of a stopped cluster is marked for permanent deletion together with its parked per-node Kubernetes record, so a start does not try to bring either back. A control-plane or bastion record, live or parked, is never retired, and a record that changed since the proof (re-pointed at another machine, resurrected, or with a job in flight on it) is left alone. The machine itself is not touched - the provider already deleted it. The Kubernetes Node object, if one lingers, is not deleted. The report lists every row by name with retired true or false and the reason. Not available for imported or managed-Kubernetes clusters, nor (yet) OVH, Scaleway, Proxmox or HPE Morpheus.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "resource_ids": "<resource_ids>"
    },
    "name": "retire_gone_machine_rows"
  }
}

Input Schema

NameRequiredDescriptionDefault
cluster_idNoUUID of the cluster. Optional when a cluster is selected.
cluster_nameNoExact name of the cluster. Optional when a cluster is selected.
resource_idsYesresource_id of each record to retire, as get_cluster_machine_inventory listed it in retirable_resource_ids.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already flag destructive=true, readOnly=false, and idempotent=false, but the description goes far beyond them: it explains the write-time proof rule (404 for the recorded machine id plus another machine of the same cluster present), the exact cascade (live worker becomes down and is permanently deleted with its per-node Kubernetes record; node group without autoscaling releases settings), the parked-cluster behavior, and that control-plane/bastion rows are never retired. Nothing here contradicts the annotations, and the race-condition carve-out (a row re-pointed, resurrected, or with a job in flight is left alone) is behavior an agent could not infer from structured fields.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose and input source are front-loaded in the first two sentences, and most of the remaining detail is substantive deletion semantics rather than filler. It is still a single very long block with layered parentheticals that could be split or trimmed, costing a point on readability.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a destructive, non-idempotent operation with no output schema, the definition covers preconditions, cascade effects, exclusions, and the shape of the return ('The report lists every row by name with retired true or false and the reason'). An agent has everything needed to call it correctly and interpret the result.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so cluster_id, cluster_name, and resource_ids are already documented at the field level, and the description mostly repeats them (the 20-per-call cap, the retirable_resource_ids provenance). It adds little new meaning for the two cluster-identification parameters, so the baseline of 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a precise verb and resource ('Retire platform server records ... whose machine the cloud provider no longer has') and immediately ties it to the sibling that produces the input, get_cluster_machine_inventory, naming the exact classifications 'phantom' and 'gone_kept'. An agent can distinguish this from inventory-read or generic retire tools without opening a schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It says exactly where the input comes from ('Pass the resource_ids it listed in retirable_resource_ids'), the batch limit, and the negative scope ('Not available for imported or managed-Kubernetes clusters, nor (yet) OVH, Scaleway, Proxmox or HPE Morpheus'). It also spells out the preconditions under which a row is left untouched (timeout, auth error, unconfirmed 404), so the agent knows when not to expect a retirement.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.