Skip to main content
Glama

Create a task

create_task

Add ONE task to a day of the user's Habidu schedule: title, date, start time, duration, optional repeat, steps (subTasks) and a "Bring / prep" checklist. Use it when the user gives a day or a time; for an undated to-do use create_inbox_item, for a step-by-step habit use create_routine, and to change an existing task use update_task. Returns the stored task with its id; it then shows in the app and gets reminders. A past date, an overlap or an identical block comes back ok:false with a hint and is NOT saved - tell the user why, never say it was added. Preview rule: if this is the second task or routine change within about 90 seconds, nothing is saved and the result is needs_preview with the pending and saved changes - show them to the user and call again with confirm:true only after they agree.

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": {
      "title": "<title>"
    },
    "name": "create_task"
  }
}

Input Schema

NameRequiredDescriptionDefault
minNoStart minute, 0-59 (default 0). With hour 14 and min 30 the task starts at 14:30.
dateNoThe task's actual calendar date as YYYY-MM-DD. Resolve any explicit date the user gives or an image shows ("June 18", "8/31", "Dec 24") to the correct day, inferring the year from the current date in context (if that month/day already passed this year, use next year unless context says otherwise). ONLY omit this for a task that is genuinely for TODAY — never default a future-dated item to today, and never put the date in the title.
hourNoStart hour, 0-23 (local). Omit for an `anytime` task (no fixed time).
titleYesShort task title, e.g. "Dentist" or "Call mom".
allDayNoTRUE for a WHOLE-DAY event that has no clock time and spans the entire day — school closures, public holidays, days off, "out of office", a birthday. It is pinned as a banner atop that day and does NOT take a time slot or duration. Use the `date` field for which day. Do NOT set `hour`, `anytime`, or `durationMin` with it. NOT for a timed appointment (a 9am doctor visit is a normal timed task).
effortNoRough effort/size: S (small, ~quick), M (medium), L (large). Omit to let it derive from the duration. Offer to split an L task into a smaller first step.
repeatNoRecurrence: Once (default, one-time), Daily (every day), Weekly (every week on the same weekday as `date`), Monthly (same day-of-month as `date`), or Yearly (same month and day as `date` every year - birthdays, anniversaries).
anytimeNoTRUE for a task with NO fixed time (a phone call, an errand) — it goes to the Inbox and stacks by `context` instead of taking a timeline slot. Set a `context` alongside it. Do NOT set both `anytime` and a specific `hour`.
confirmNoSet true ONLY after the user has seen and approved this change in a preview you showed them (needed when a request changes more than one task or routine). Never set it on your own. Default false.
contextNoBatch/context tag for stacking same-context tasks: calls (phone calls), commute (do while commuting), errands (out-and-about), deep (focused work sessions ONLY - never a habit or health nudge like drinking water, which is not deep work and must not stack with it). Use it for tasks that group naturally — most phone calls are `calls`.
endDateNoOnly for a MULTI-DAY all-day event (allDay:true): the LAST day of the span as YYYY-MM-DD (inclusive), with `date` as the first day. E.g. "school is closed all week" → date = Monday, endDate = Friday. Omit for a single-day event.
categoryNoOptional category label, e.g. Work, Personal, Health, Learning, Creative, Errands, or one of the user's own.
priorityNoPriority: 4 Very High, 3 High, 2 Medium, 1 Low. Set ONLY when the user expresses importance/urgency ("this is critical", "top priority"). Omit otherwise.
subTasksNoOptional ordered steps, e.g. [{"name":"Warm up","durationMin":10}].
checklistNoA "Bring / prep" checklist the user ticks off: things to bring or have ready ("bring the contract and my laptop to the meeting" -> ["Contract", "Laptop"]). Short items, each its own entry. NOT steps (those are timed parts of the block) and never put them in the description.
day_timesNoA DIFFERENT start time on some weekdays of a REPEATING task. `hour`/`min` are the usual time; list only the weekdays that differ. "5:15 on Tuesdays and Thursdays, 7:30 the other days" -> hour 7, min 30, day_times [{weekday:2,hour:5,minute:15},{weekday:4,hour:5,minute:15}]. ONE task, never two. Omit when every day is the same.
alert_typeNoHow the reminder fires: "alarm" = a true alarm (rings until dismissed), "push" = a plain notification. Set ONLY when the user explicitly asks ("wake me with an alarm", "just a quiet reminder"). Omit = their default.
daysOfWeekNoSPECIFIC weekdays the task repeats on: 1=Monday .. 7=Sunday. Use when the user names days — "Monday to Friday"/"weekdays" → [1,2,3,4,5]; "weekends" → [6,7]; "every Mon, Wed, Fri" → [1,3,5]. Overrides the same-weekday-as-date default (set repeat:"Weekly" alongside; the anchor `date` may be any day). Never invent days the user did not state; omit entirely for plain cadences.
descriptionNoExtra detail that belongs WITH the task but not in its title: a phone number, an address, an account or booking reference, a note to self (things to BRING or have ready go on `checklist`, not here). Use it instead of padding the title - "Call the plumber" with description "555-1234, ask about the leak under the sink". Omit when there is nothing extra to carry.
durationMinNoDuration in minutes (defaults to 30 when omitted)
in_progressNoSet TRUE when the user says they are CURRENTLY doing this ("I'm working 1 to 8:30 pm", "I'm in a meeting till 6"): the block is kept on today at its STATED start and duration and marked started, even if it began only minutes ago. Never set it for something they merely plan to do.
contact_nameNoWho a CALL task is to (person/provider), e.g. "Dr. Smith" / "the plumber".
delegated_toNoThe NAME of the person who will do this task instead of the user ("my wife will do the pickup" -> "Wife"). A delegated task stays on the schedule, never counts as an overlap and gets no start reminder - one "Did <name> handle it?" check after it ends. Only when the user says someone else covers it. "Assindo" is the built-in delegate - the user's AI phone assistant app: it makes the phone call or posts to social media for this task. Use delegated_to "Assindo" only when the user asks Assindo to do it; Habidu sends it the task when it is saved (it needs Assindo connected in Settings - Connected apps).
skip_routineNoONLY for a one-day wake-up or lie-in that REPLACES that day's routine ("usually 5, but tomorrow 8"): skips the named routine on this task's date so it does not also ring. Never for an ordinary task. One of: morning | evening.
alarm_missionNoA MISSION that must be done before this ALARM stops ringing (Android and iPhone): "math" = solve math problems, "shake" = shake the phone, "steps" = walk a number of steps, "qr" = scan a QR code / barcode the user registered in Settings > Reminders & alarms > Mission codes. "none" removes it. Setting a mission makes the task an alarm. Set it ONLY when the user explicitly asks ("make me solve math to turn it off", "an alarm I have to shake to stop", "make me walk 200 steps"). Never unprompted.
allow_overlapNoSet TRUE only after the user confirmed they want the task despite a scheduling conflict you already warned them about — skips the overlap check.
contact_phoneNoThe phone number for a CALL task, as given. NEVER invent one — omit if the user has not provided it.
alarm_mission_countNoHow many problems (math, 1-10, default 3), shakes (shake, 10-100, default 30) or steps (steps, 20-1000, default 100). Only with alarm_mission.
alarm_mission_levelNoMath difficulty: easy | medium (default) | hard. Only with alarm_mission "math".

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.8/5.0
Behavior5/5

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

Annotations only declare the generic profile (readOnlyHint=false, destructiveHint=false, idempotentHint=false), so the description carries the real burden and does: it discloses the return value (stored task with id, appears in app, gets reminders), the three rejection cases (past date, overlap, identical block -> ok:false, NOT saved) with an instruction to tell the user why, and the needs_preview/confirm gating rule. That is behavioral detail no annotation or schema field conveys.

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?

Front-loaded with purpose, then routing, then failure semantics, then the preview rule — a sensible order for a 29-parameter tool. It is dense but nearly every clause earns its place; the only compression gain would be tightening the preview-rule sentence.

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 29-parameter mutation tool with no output schema, the description still covers the success return (task with id), the failure return shape and its user-facing handling, and the multi-change preview protocol with the `pending`/`saved` payload. An agent has enough to call it correctly and to report results honestly without further documentation.

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% and the per-field descriptions are already extensive, so baseline would be 3. The description adds genuine routing meaning by tying the date/anytime decision to create_inbox_item ('for an undated to-do use create_inbox_item'), which is the one parameter distinction the schema alone doesn't arbitrate; it does not add syntax-level detail beyond the schema.

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?

States a specific verb and resource with scope ('Add ONE task to a day of the user's Habidu schedule') and enumerates the content it carries (title, date, start time, duration, repeat, steps, checklist). It explicitly contrasts itself with three siblings (create_inbox_item, create_routine, update_task), so an agent can route without opening another 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?

Gives an explicit selection rule ('Use it when the user gives a day or a time') plus the exact alternatives for the two adjacent cases: undated to-dos and step-by-step habits, and update_task for edits. It also states a when-not path via the preview rule ('call again with confirm:true only after they agree').

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.

Resources