Skip to main content
Glama

Propose off hours schedule

propose_off_hours_schedule
Destructive

Propose stopping a cluster's worker servers outside working hours to save cost: the off-hours lever of the Cost decisions, filed as an off_hours_schedule decision a person approves, never written straight to the cluster's power schedules. Prefer it over create_power_schedule whenever the user wants a cluster off at night or at weekends to save money, or acts on an off_hours_schedule recommendation from get_cloud_savings. The stop is scale_to_zero (the worker servers are deleted at the provider every stop and created again at the start; the control plane, etcd and cloud volumes keep running, and anything on a worker's own disk is lost at every stop). It runs the same off-hours gate the savings model and the cost autopilot run: a cluster nobody confirmed as non-production, one labelled or graded production, one that served traffic or ran its own workloads inside the stop window over the last 14 days, or one whose activity could not be measured is refused with the reason and nothing is filed - tell the user why, and do not schedule the stop with create_power_schedule instead. When another surface (the cost autopilot, the Cost page) already holds an off-hours proposal for the cluster, that proposal is returned (standing_twin true) and no second one is filed. On the confirmation card, the member who confirms it approves the proposal and Ankra arms the stop and start schedules at once, with a receipt, and measures the saving seven days later; that needs billing.manage. Everywhere else - MCP, an approved plan - the tool only files the proposal and returns portal_url, where a person with billing.manage approves it: give the user that link. The default window is weekdays 19:00 to 07:00 plus weekends, in the organisation's off-hours timezone; the expected saving is recorded only for that window. A window the user chose (stop_cron, start_cron) is measured over the hours it stops, not the default window's. A proposal already approved, held or running for the cluster keeps the window a person approved: a request for another window is refused with both windows (requested_window), and a person sets the standing proposal aside at portal_url before a new one is filed. Set accept_node_local_data_loss true ONLY after the user has explicitly agreed to lose the data on worker disks at every stop, never on your own judgement.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "propose_off_hours_schedule"
  }
}

Input Schema

NameRequiredDescriptionDefault
timezoneNoIANA timezone the stop and start run in, e.g. 'Europe/Stockholm'. Defaults to the organisation's off-hours timezone (the cost autopilot's quiet hours, else UTC); the cluster's night activity is measured in this timezone.
stop_cronNo5-field cron of the stop. Defaults to '0 19 * * 1-5' (weekdays 19:00). Only change it when the user asks for another window.
cluster_idNoUUID of the target cluster. Optional when a cluster is selected.
start_cronNo5-field cron of the start. Defaults to '0 7 * * 1-5' (weekdays 07:00). Must never fire at the same time as the stop.
cluster_nameNoExact name of the target cluster. Optional when a cluster is selected.
accept_node_local_data_lossNoThe user's explicit consent to losing the data on the worker disks at every stop, used when the card's approval runs the schedules. Set it to true ONLY after you have told the user what is lost and the user has explicitly agreed - never pre-emptively, and not because the user asked for the schedule. Defaults to false.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A4.7/5.0
Behavior5/5

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

Beyond the destructiveHint=true annotation, it explains precisely what is destroyed ('worker servers are deleted at the provider every stop ... anything on a worker's own disk is lost at every stop') while noting the control plane, etcd and volumes persist. It also discloses the gate criteria, twin-proposal behavior, billing.manage permission requirement, and the seven-day measurement delay.

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

Conciseness3/5

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

Purpose is front-loaded and nearly every sentence carries operational meaning, but the entire definition is a single dense paragraph mixing gating, approval, window semantics and consent. Sub-structuring or trimming lower-value restatements would improve scanability without losing content.

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, gated, approval-flow tool with no output schema, the description covers the data-loss consequence, gate refusal reasons, twin proposals, approval paths and portal_url, permission requirements, and window conflict handling. An agent has everything needed to use it safely and route the user correctly.

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

Parameters4/5

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

Schema coverage is 100%, so the baseline is 3, but the description adds meaning: the savings for a user-chosen window are measured over the hours it stops rather than the default window, and conflict handling for an already-approved window. It reinforces the strict consent rule for accept_node_local_data_loss, though much of that is already in the schema text.

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 first sentence gives a specific verb+resource+'why' ('Propose stopping a cluster's worker servers outside working hours to save cost') and immediately frames it as the off-hours lever of Cost decisions. It explicitly contrasts with create_power_schedule, so the agent can separate it from its closest sibling.

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 states exactly when to prefer it ('whenever the user wants a cluster off at night or at weekends to save money, or acts on an off_hours_schedule recommendation from get_cloud_savings'), when not to fall back to create_power_schedule, and what to do when the off-hours gate refuses. That is explicit when/when-not/alternative guidance.

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.