MCP App Store
by KeenEthics
VaktFlyt icon

VaktFlyt MCP Server for Claude and ChatGPT

By Planless AS

Overview

VaktFlyt is a Norwegian workforce-management service for shift scheduling (turnus), absence, shift swaps and compliance with the Norwegian Working Environment Act. This connector lets a VaktFlyt user work with their own organisation's data from Claude. The connector acts as the signed-in user. It sees only what that person can see in VaktFlyt, and the tools it offers follow the person's role (employee, leader, administrator, union representative, data protection officer). Read tools cover rosters and shifts, employees, organisation and departments, staffing and open shifts, absence and shift swaps, reports on hours and overtime, checks against the Working Environment Act chapter 10, and look-ups in law, collective agreements and VaktFlyt's help texts. Propose tools never change data. Each call adds to one plan and returns a dry run of the whole plan and an approval link. The user opens the link and approves or discards the plan in VaktFlyt. Approved changes pass the same permission and working-time checks as changes made in the app. Privacy: information under GDPR Article 9, such as diagnosis and reason for absence, is removed before any answer leaves VaktFlyt. Every call is recorded in the organisation's audit trail. The service is hosted in the EU. Requirements: a VaktFlyt organisation on the Profesjonell or Bedrift plan, where an administrator has turned on AI clients (off by default) and recorded the two confirmations Norwegian and EU law require before AI is used at work. Each user then signs in with their own VaktFlyt account and chooses read-only access or read and propose. Tool titles and most answers are in Norwegian, because VaktFlyt's users are. Claude answers in the user's language.

How to connect

  1. Claude: Open the VaktFlyt listing in the Claude connectors directory and choose Connect. Claude listing VaktFlyt exposes 17 tools here.
  2. ChatGPT: Open the VaktFlyt listing in the ChatGPT app directory and choose Connect. ChatGPT listing VaktFlyt exposes 17 tools here.

Tools

absence_and_swaps

ChatGPT
Absence, holiday wishes, approvals and substitute searches (cascades). Operations ( = required for that operation): - my_absences: The caller's own absences. (from_date, to_date) - my_quotas: The caller's leave balances. (year, leave_type) - my_vacation_wishes: The caller's own holiday wishes (ferieønsker) for a year: periods, rank, note, status, and the virkedager the year holds. (year) - vacation_round: The department's vacation round (ferierunde) for a year: deadline, sent out or not, who has answered and who is missing, and the vacation plan's status. Leader. (department_id, year) - vacation_balances: Ferie days left. employee_id: one person (used, pending, carried over, senior days, feriepengesats, carryover review). department_id: everyone, with who has most left, none taken, above the carryover ceiling. Leader. (employee_id, department_id, year) - vacation_board: The ferietavle: each person's wishes (rank, weeks), who has not answered, the weeks in conflict (first choices vs floor, as the board in the app grades them), plan status. Answers «how many are away in week 30 if everyone gets their first choice». Leader. (department_id, year) - vacation_outcome: Why an employee got or missed the ferie they asked for: reason, queue position, debt, which ranks were also full; the leader also gets the latest proposal. Employee: own. (employee_id, year) - away: Who is away each day (≤31 days): name, ferie/fravær, godkjent/søkt. Never why. (department_id, from_date, to_date, group_by) - pending_approvals: Absences, swaps, timesheets and single vacation wishes (enkeltønsker) waiting for the leader's decision, with names, dates and ids (timesheets: date, times and hours). (department_id, item_type) - sla_breaches: Godkjenninger › Fristbrudd: cases answered later than the answer target, with the total, how many still wait, and why the log cannot be cleared. (department_id, from_date, to_date) - flyt_cases: What Flyt itself has done, case by case with a timeline: a freed shift and its substitute search, a wish round, a draft turnus, a fixed-rhythm question. case_id: opens that one case's page in the app. Leader. (case_id, department_id) - cascade_status: Status of one substitute search. (swap_id) - active_cascades: Substitute searches in progress. (department_id, include_resolved, include_stats, days) - my_cascade_status: Substitute searches for the caller's own absence and swap requests (my_role), and the ones waiting for the caller's answer (asked_of_you). (absence_id) - cascade_ranking: Why candidates were ranked as they were. (swap_id, candidate_rank)

compliance

ChatGPT
Working Environment Act (arbeidsmiljøloven) checks. Binding results come from these checks, not from memory. Operations ( = required for that operation): - validate_schedule: All violations in a schedule. (schedule_id) - check_shift: Whether a planned shift for an employee breaks a rule. (employee_id, date, start_time, end_time, department_id, also_planned) - violations: Violations, grouped. Needs from_date, to_date and department_id or schedule_id. (from_date, to_date, department_id, schedule_id) - rest_between: Rest time between two shifts. (shift_a_id, shift_b_id) - limits: Statute limits (hours, rest, nights); organisation_limits is what this org runs. - rule: One provision of the Work Environment Act ch. 10 with its limits (not ferieloven; use knowledge search for that). (section*, include_tariff_variants) - expiring_certifications: Employees' certificates expiring within a window, and expired ones. (window_days, include_expired, department_id)

knowledge

ChatGPT
Labour law, tariff agreements, VaktFlyt documentation and the organization's own uploaded documents (handbooks, local agreements, routines; scope: organization), as cited passages. Search before answering anything about what is allowed, required or paid, even when you feel sure, and cite what you used. Passages come best first (rank); truncated: true means the source continues beyond the text shown. Operations (* = required for that operation): - search: Passages matching a question, with citations. - rule: One provision by reference, e.g. 'aml § 10-8 (1)'.

my_account

ChatGPT
The caller's profile, time entries, confirmations, wish rounds, privacy status and onboarding. Operations ( = required for that operation): - profile: The caller's own registered profile. - availability_preferences: The caller's own availability preferences (nights, weekends, days off). (preferences, department_id) - time_entries: The caller's clock-in/out entries. (from_date, to_date) - unconfirmed_shifts: Worked shifts the caller has not confirmed. - privacy_status: The caller's data-protection requests and status. - explain_feature: How a VaktFlyt feature works. (feature_name, category)

navigate

ChatGPT
Get a link to the VaktFlyt page where the user does something themselves: registering absence, regenerating a roster, payroll export, privacy requests, breach reporting, or any page. The answer holds a url the user opens in a browser; nothing opens by itself. Use it when a change cannot be proposed here. Operations ( = required for that operation): - page: Link to a page, optionally with steps to follow there. (page, steps, params) - register_absence: Link to absence registration. A leader or admin lands on registering it for an employee on the approval board (pass employee_id, start_date and end_date to open it filled in, never the type or reason); own_absence links their own absence page. (own_absence, employee_id, start_date, end_date) - regenerate_schedule: Link to regenerating a roster with a changed constraint. (schedule_id, constraint, value) - payroll_export: Link to the payroll export. (from_date, to_date, department_id) - compliance_report_export: Link to the compliance report for a week, to download as CSV or Excel. (week_date, department_id) - request_erasure: Link to the erasure request. - report_breach: Link to breach reporting.

organisation

ChatGPT
Departments, positions, qualifications and the organisation's configuration. Operations ( = required for that operation): - find_departments: Departments by name; returns ids for other calls. (query, limit) - positions: Positions (stillinger) defined in the organisation, with ids and how many hold each. - position_usage: Where one position is used: shift types and departments that require it (counts), upcoming shifts per department with the first dates, templates, profiles, rules and who holds it. Read it before changing, merging or deleting a position. A leader sees the departments they lead. (position_id, name) - shift_types: The organisation's own shift types: id, code, name, default times, break, category, departments and position requirements. - certification_types: Certification types required by shift types. - qualification_substitutions: Which qualifications may substitute for which. - industry_config: The sector's default setup; the organisation's own shift types are shift_types. - knowledge_documents: Titles of every document the knowledge tool searches: the shared corpus and the organisation's uploads. To read one, search the knowledge tool; no list needed. (document_type, page, page_size) - notification_templates: Notification templates. (active_only) - webhooks: Configured webhooks. (active_only) - read_attached_file: A file the user attached («Vedlagt fil: fil_…»): counts and codes, never names or content. Saves nothing. with_ai only after needs_ai and the user's yes. (file_handle, with_ai) - assemble_attached_files: Read files for one department. (file_handles*, department_id, department_name)

oversight

ChatGPT
Payroll readiness and the audit trail (admin, payroll and data-protection roles). Operations ( = required for that operation): - payroll_readiness: Whether a period is ready for payroll export. (from_date, to_date, department_id) - payroll_preview: What a payroll export would contain. (from_date, to_date*, department_id) - audit_search: Search the audit log. (q, action, entity_type, entity_id, from_date, to_date) - audit_stats: Audit log statistics. - breach_incidents: Registered personal-data breaches. (status) - processing_records: Records of processing activities (opens the page).

people

ChatGPT
Look up employees and their load. Results name people the caller may see. Operations ( = required for that operation): - find: Find employees by name, department or role. (query, department_id, limit) - availability: An employee's availability and absences in a period. (employee_id, from_date, to_date) - workload: Hours and shifts of one employee in a period, against contract. (employee_id, from_date, to_date) - weekend_load: Weekend shifts of one employee in a period. (employee_id, from_date, to_date) - work_pattern: One employee's work pattern (which weekdays, how often, only those days or not) and the upcoming shifts it blocks. (employee_id*)

plan

ChatGPT
The plan this token has proposed (every propose_ call adds to one plan that the user approves in VaktFlyt). Use it to see what became of the plan after you gave the user the approval_url, and to get out of a plan that cannot be approved. Nothing here changes data. Operations ( = required for that operation): - status: Whether the plan is waiting, running (approved, its roster draft being made), approved, undone, failed, discarded or expired, each change's outcome, and whether it can be undone, per change and as a whole; undo_url is where the user undoes an approved plan (changeset_id; the newest plan when left out) - drop: Take changes out of the waiting plan; the rest is checked again. Needs the call_ids to remove and the plan_token of the plan you read them from (call_ids, plan_token, changeset_id) - discard: Close the waiting plan without running it; the next proposal starts a new plan (changeset_id)

planning

ChatGPT
Scheduling profile, feasibility and explanations for roster planning. Operations ( = required for that operation): - profile: A department's scheduling profile (rules for generation). (department_id) - team_wishes: Team wishes: no nights/weekends, more work etc. (department_id, topic) - feasibility: Whether a period can be staffed at all. (department_id, from_date, to_date) - feasibility_preview: Feasibility for a selection of employees. (department_id, from_date, to_date, selected_employee_ids) - preference_feasibility: Whether stated preferences can be met. (department_id, from_date, to_date, preferences) - topic_plan: Planning topics still open before generation. (department_id, from_date, to_date, schedule_id) - explain_employee: Why an employee got the shifts they got. (schedule_id, employee_name, shift_type, date, weekday) - explain_concept: A scheduling concept in plain words. (concept) - suggest_improvements: Schedule improvements; with employee_id, kind, count: same-day swaps off one person. (schedule_id, employee_id, kind, count)

propose_my_changes

ChatGPT
Propose changes to the caller's own shifts, time, profile, holiday wishes and wish-round answers. Added to the plan the caller approves. Operations ( = required for that operation): - claim_open_shift: Take an open shift. (shift_id, overtime_consent) - request_swap: Ask colleagues to take one of your shifts. (shift_id, candidate_ids, candidate_limit, reason, notes) - confirm_shift: Confirm a worked shift. (shift_id, extra_minutes) - clock_in: Clock in. (shift_id, notes) - clock_out: Clock out. (entry_id, notes) - update_profile: Change your name or phone. (first_name, last_name, phone) - notification_preferences: Change how you are notified. (email_enabled, sms_enabled, push_enabled, quiet_hours_start, quiet_hours_end) - availability_preferences: Set your availability preferences. (preferences, department_id) - request_data_export: A signed export of the personal data stored about you (GDPR art. 15/20), made on approval with a one-time download link. Ends the plan. - cancel_absence: Cancel a registered absence. (absence_id) - submit_vacation_wishes: Send your holiday wishes for a year to the leader («Send ferieønsker»). Takes the WHOLE set and replaces what was sent: to add, change or remove a wish, read my_vacation_wishes and send the set as it should be. (year, periods, within_hovedferie, preference_id) - withdraw_vacation_wishes: Withdraw all your holiday wishes for a year, before the leader locks them. (year, preference_id) Ends the plan: request_data_export. At most one such operation per plan, and it must be proposed last, after every other change; a plan that already holds one takes no more proposals until the user approves or discards it in VaktFlyt.

propose_org_changes

ChatGPT
Propose organisation setup changes. Added to the plan the approver approves. Propose all changes of a clean-up in the same response; they are checked together. Operations ( = required for that operation): - create_department: Create a department. (name, description) - update_department: Rename or change a department. (department_id, name, code, description, manager_id, min_staffing) - deactivate_department: Deactivate a department. (department_id) - reactivate_department: Reactivate a department. (department_id) - deactivate_departments: Deactivate several departments. (department_ids) - create_position: Create a position. (name, description) - update_position: Rename or change a position (name, description, code, hourly rate, baseline certifications). A rename carries every requirement, template, profile, rule and upcoming shift with it, merging where the new name is already required. A name another active position has is refused: merge with delete_position and replace_with. The dry run counts what moves. (position_id, name, description, code, hourly_rate_nok, baseline_certifications) - delete_position: Delete one position, or several via position_ids (all or none). Refused while a requirement or an upcoming shift names one, unless replace_with (another position takes its requirements, upcoming shifts and holders: a merge) or strip_requirements (the requirements go; fixed crews get a warning). The refusal names every blocked one. The dry run counts what moves or goes. (position_id, position_ids, replace_with, strip_requirements) - set_member_position: Give employees a position. (user_ids, position_id) - move_members: Move employees between departments. (from_department_id, to_department_id, user_ids, position_id, reason) - invite_member: Invite a person to the organisation. (email, role, department_id, invitations) - create_shift_types: Create shift types. (shift_types) - delete_shift_type: Delete a shift type. (shift_type_id) - company_setup: Set up departments and positions from a sector template. Additive: it only creates what is missing, so never use it to clean up, rename or remove. (sector, department_names, position_names, create_default_departments, create_default_positions, create_default_shift_types, skip_existing, dry_run) - save_setup: Set up one department for a new customer in ONE plan: the people from attached files (file_handles, read first with organisation.read_attached_file) and/or the people the user typed (people), with positions, shift types (own_shift_types, or the industry's usual ones), the staffing and the position seats the industry asks for. People are saved as placeholders without e-mail and nobody is invited. The dry run is the whole save: plan_lines say what approving leaves behind. Names in a file never reach you. (file_handles, department_id, department_name, people, own_shift_types) Ends the plan: invite_member. At most one such operation per plan, and it must be proposed last, after every other change; a plan that already holds one takes no more proposals until the user approves or discards it in VaktFlyt.

propose_planning_changes

ChatGPT
Propose roster-planning decisions: staffing demand, work patterns, the planning profile, a roster draft, publishing, decisions on requests and the vacation round. Added to the plan the leader approves. Operations ( = required for that operation): - draft_roster: Generate an unpublished roster draft for a department and period from its saved profile. The dry run is the generation's preflight (shift types, people, shifts needed, warnings, AML blockers); the draft is made only on approval. It ends the plan: one per plan, proposed last. (department_id, from_date, to_date, period_notes, include_employee_ids) - set_staffing: Set how many one shift type needs in a department, as in Oppsett: every_day, weekdays and/or weekend, or by_weekday; days not named keep today's number. The dry run shows the demand before and after per day. Put it before a draft_roster in the same plan and the draft is planned against it. (department_id, shift_type_id, every_day, weekdays, weekend, by_weekday) - set_work_pattern: Set an employee's whole work pattern: groups of weekdays, each with its cadence (every week, every other week), and only_these_days. The dry run says it in words and lists the upcoming shifts it blocks; add propose_shift_changes.unassign for each of those in the same plan. (employee_id, groups, only_these_days, weekend_nei_answer) - add_employee_rule: Save a lasting rule for one employee that every later turnus keeps: EITHER weekdays OR shift types (day, evening, night) they are not planned on, until valid_to or until removed. Never pass or repeat the reason. The dry run says the rule back and lists upcoming shifts it covers. (employee_id, days, shift_types, valid_from, valid_to) - update_profile: Change a department's scheduling profile. (department_id, changes) - union_consultation: Record the union's § 10-3 review of a schedule (outcome, comments, and the consultation_round_id from schedule.consultation_status). (schedule_id, outcome, comments, consultation_round_id) - publish: Publish a consultation-complete schedule to employees. The dry run runs every publish check (consultation, AML, § 10-3 notice, overlap) and reports what would block; employees are told on approval. It ends the plan: one per plan, proposed last. (schedule_id, notice_override_reason, sector_rule_override_reason, competence_override) - unpublish: Take a published schedule back to draft. The dry run checks it can be retracted; on approval employees are told and open swaps and claims on it end. It ends the plan: one per plan, proposed last. (schedule_id) - decide_request: Approve or reject a pending absence, swap, open-shift claim, timesheet or single vacation wish (item_type absence, swap, open_claim, timesheet or vacation_wish; ids from absence_and_swaps.pending_approvals; one item_id, or several via item_ids). open_claim: a claim on an open shift; a claim or swap in a published or consultation-locked turnus is refused and decided on Godkjenninger. The dry run is the whole decision, rolled back; on approval the people concerned are told and an approved absence starts its cover search. Refused, for the leader: AML or overtime overrides, a vacation_wish in the notice period (ferieloven § 8 nr. 1). (item_id, item_type, action, item_ids, reason) - revoke_absence_approval: Take back an APPROVED absence, as «Trekk tilbake godkjenning» in the app: its shifts go back to the employee where the Act allows, the leave is credited, and on approval the employee is told. Colleagues the cover search already asked have had their message. A pending absence is rejected with decide_request instead. (absence_id, reason) - set_vacation_round_deadline: Set the deadline for holiday wishes (ferieønsker) in a department and year. Advice, not a lock: a late wish is still accepted. (department_id, year, deadline_date) - register_vacation_wish: Register the whole holiday-wish set for an employee who cannot themself (no app, a paper answer), marked registered by the leader. Refused when the employee sent their own. (department_id, year, employee_id, periods) - o…

propose_shift_changes

ChatGPT
Propose changes to shifts. Nothing changes now: every call adds to the plan the leader approves, and returns a dry run of the whole plan (including rule violations). Send all changes of one kind together: several in one call where the operation takes a list (shifts, assignments), otherwise one call each in the same response; they are checked together, and the plan accumulates. Operations ( = required for that operation): - create: Create one shift, or several via 'shifts'. (department_id, date, start_time, end_time, employee_id, aml_override_reason, shifts) - move: Move a shift to another date or time. (shift_id, new_date, new_start_time, new_end_time, aml_override_reason) - assign: Assign an employee to an open shift. (shift_id, employee_id, overtime_consent_attestation, aml_override_reason, assignments) - unassign: Remove the employee from a shift. (shift_id) - delete: Delete a shift. (shift_id) - cancel: Cancel shifts of a type on a date. (schedule_id, date, shift_type) - swap: Swap the employees of two shifts. (shift_id_a, shift_id_b, aml_override_reason) - fill_open_shifts: Fill open shifts with eligible employees. (department_id, schedule_id, from_date, to_date, shift_ids, not_employee_ids)

reports

ChatGPT
Hours, overtime, absence, fairness and cost. Operations ( = required for that operation): - hours: Hours per employee: entries or planned shifts. (department_id, from_date, to_date, source) - overtime: Overtime per employee, from time entries. (department_id, from_date, to_date) - merarbeid: Additional hours for part-time employees. - absence: Absence per type in a period; names nobody. (department_id, from_date, to_date) - fairness: Fairness of nights, weekends and public holidays (helligdager), not vacation. (department_id, from_date, to_date) - fairness_over_periods: A department's fairness indices in two periods, and the change. (department_id, first_from_date, first_to_date, second_from_date, second_to_date) - fairness_gaps: Who carries too much or too little in a schedule. (schedule_id, dimension) - cost: Wage cost and tariff supplements of a schedule. (schedule_id, compare_schedule_id) - compare: Differences between two schedules. (schedule_id_a, schedule_id_b)

schedule

ChatGPT
Read rosters and shifts. Operations ( = required for that operation): - department_shifts: Department shifts (`employee_id`: one person's, `only_open`); `aggregation` counts. (department_id, from_date, to_date, schedule_id, include_drafts, employee_id, only_open) - my_shifts: The caller's own shifts for a period. (from_date, to_date) - list_schedules: Schedules (rosters) of a department, with status. (department_id, status) - fill_status: How much of a schedule is staffed, and what is open. (schedule_id) - view_summary: Summary of one schedule: status, period, shifts (total, assigned, open) and the mix of shift types. (schedule_id) - generation_context: Why a schedule is as it is: inputs, solver path, warnings, open shifts by cause. (schedule_id) - consultation_status: Where a schedule's union consultation (aml § 10-3) stands, with the consultation_round_id a review must name. (schedule_id) - consultation_package: A schedule summarised for the union's § 10-3 review: shifts, coverage, overtime and the publication notice. (schedule_id*)

staffing

ChatGPT
Staffing demand, coverage, open shifts and who can work. Operations ( = required for that operation): - demand: How many each shift type of a department needs per weekday (the demand rosters are planned against, as Oppsett shows it), and which numbers are saved. (department_id) - week_status: Day-by-day staffing status of a department. (department_id, from_date, days) - levels: Staffing against minimum staffing for a date or period. (department_id, date, to_date) - open_shifts: Open (unassigned) shifts. (department_id, from_date, to_date, limit) - substitutes: The best substitutes for a shift someone is away from, ranked: employee_id and date in ONE call (finds the shift, says if the absence is registered), or shift, or date and time of day. Contacts no one. Open shift: available_employees. (shift_id, date, time_of_day, employee_id, department_id, limit) - position_requirements: How many of each position a department's shift types require (its per-position matrix), and the requirements its shift types carry themselves. (department_id) - available_employees: Employees free for a given time (no shift in any department, no absence) whom the shift would put in breach of no working-time rule; those it would are listed under blocked_by_aml with the rule. (department_id, date, start_time, end_time) - eligibility: Whether one employee can take one shift, and why not. (employee, date, time_of_day, department_id, shift_id) - absence_forecast: Expected absence in a period. (department_id, from_date, to_date) - risk_radar: Upcoming staffing and compliance risks. (department_id, days_ahead, from_date, to_date) - what_if: Staffing effect of a hypothetical change. (department_id, from_date, to_date, scenario*)

absence_and_swaps

Claude

compliance

Claude

knowledge

Claude

my_account

Claude

navigate

Claude

organisation

Claude

oversight

Claude

people

Claude

plan

Claude

planning

Claude

propose_my_changes

Claude

propose_org_changes

Claude

propose_planning_changes

Claude

propose_shift_changes

Claude

reports

Claude

schedule

Claude

staffing

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/vaktflyt"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/vaktflyt)