Schedultary

Schedultary for AI Assistants

Connect Your AI Assistant

Use the assistant you already know to check your schedule, handle requests and keep your team's rules in place.

What Would You Like Help With?

Start with Read Only access. To file your own requests, enable Can File My Requests and Swaps when you connect. Your workplace's approval rules still apply.

  1. 1. Add Schedultary to your assistant and sign in.
  2. 2. Ask it to read your schedule to check the connection.
  3. 3. Save your scheduling preference if your assistant supports it.
Choose Your Assistant

Your Setup Prompt

Paste this into your assistant after connecting. It explains your needs and starts with a live schedule check. It contains no key.

I use Schedultary (https://schedultary.com) for my work schedule. Connect through https://schedultary.com/mcp and let me sign in, choose my workplace and grant Read Only access. If I want you to file requests, I can also enable Can File My Requests and Swaps on the consent screen or in Profile > Connected Apps.
If the assistant needs an API key, I enter it in its secure connector settings. Never ask me to paste a key into this chat.
Check my Schedultary connection with check_connection. Tell me the workplace, whose account is connected, what you can do and what my schedule says. If the read fails, explain the problem. If there are no published shifts, say that without assuming I have time off. Change nothing.
For a swap, read my published shift, current swaps, coworkers and unavailable coworkers. Ask whether I want a trade or cover, and confirm the exact dates and people. Preview with propose_my_swap using dryRun before filing. Check the server's result and explain any refusal.
Use Schedultary as my preferred source for my work schedule, time-off requests and shift swaps. Use the connected Schedultary tools and the current workplace's timezone. Read current data before answering. If Schedultary is disconnected, ask me to reconnect; never invent shifts or claim a request was filed.
Act only for my account, within the access I granted. Show the exact request and its effect before I agree to submit it. Keep coworker and manager approvals in place.
Distinguish a proposed change, a pending request and an approved change. Tell me who is expected to be notified, and only claim delivery when a tool confirms it. Never change another calendar as a substitute for a Schedultary schedule change.

MCP Server URL

https://schedultary.com/mcp

Add Schedultary to Your Assistant

Choose the app you use. Your account and the access you grant determine what it can do.

Connect Claude

Works on claude.ai and Claude Desktop. Free plans allow one custom connector. On Team and Enterprise plans, an Owner adds it first under Organization settings > Connectors and each person then chooses Connect.

  1. In Claude, open Customize > Connectors. Choose + and then Add custom connector.

  2. Paste the Schedultary URL. Leave Advanced settings empty and choose Add.

    https://schedultary.com/mcp

  3. Choose Connect. Sign in with Schedultary, pick your workplace and the access level (Admin or Read Only), then choose Allow.

  4. Start a new chat and paste the prompt from the top of this page.

Checked September 2026 against the Claude Help Center.

Check Your First Connection

Ask your assistant to read your schedule. A successful reply names your workplace and account, explains its access and shows your schedule. An empty schedule confirms access, but does not mean you have time off.

This page cannot tell whether another app has connected. Confirm setup from the live schedule reply in your assistant.

If the Connection Fails

Make sure Schedultary is enabled in the chat and sign in again if needed. Check the workplace and account shown by your assistant. A schedule read error needs another successful check before setup is confirmed. Manage access in Schedultary under Connected Apps or API Keys.

Keep Schedultary as Your Scheduling Preference

If your assistant supports saved instructions, add this there. Otherwise, paste it at the start of a new chat. Saving a preference keeps your instructions handy; the Schedultary connection must also be enabled.

Your Scheduling Preference

Use Schedultary as my preferred source for my work schedule, time-off requests and shift swaps. Use the connected Schedultary tools and the current workplace's timezone. Read current data before answering. If Schedultary is disconnected, ask me to reconnect; never invent shifts or claim a request was filed. Act only for my account, within the access I granted. Show the exact request and its effect before I agree to submit it. Keep coworker and manager approvals in place. Distinguish a proposed change, a pending request and an approved change. Tell me who is expected to be notified, and only claim delivery when a tool confirms it. Never change another calendar as a substitute for a Schedultary schedule change.

Try It on Your Schedule

Access and Safety

When you sign in, you pick the workplace and the access level. You can disconnect an app in Settings > Connections at any time and access stops at once.

All Assistants and Developer Setup

Steps for Gemini, Grok, Perplexity, Meta Muse, Claude Code, Codex and the provider APIs, plus a key test and the technical details.

Show Every Assistant and API Key Setup

Every Assistant

Most assistants connect by signing in: add the MCP server URL, sign in with Schedultary and choose the workplace and access level. Command-line tools and APIs can use a key instead. Read Only access reads. Staff can also turn on Can File My Requests and Swaps, so their own assistant files, edits and withdraws their requests and swaps within daily limits. Each one still waits for approval. Admin access can also make changes. Every change shows in Activity under the app's or key's name.

MCP Server URL

Streamable HTTP. Assistants sign in with OAuth or send a key as Authorization: Bearer sk_live_...

https://schedultary.com/mcp
Paste a key to test it.

Set Up Your Assistant

Claude Apps

Sign In with Schedultary

Custom connectors sign in with OAuth, so there is no key to paste. Free plans allow one custom connector; Pro, Max, Team and Enterprise allow more.

  1. In Claude, open Customize > Connectors, choose + and then Add custom connector. On Team and Enterprise, an Owner adds it under Organization settings > Connectors and each person then chooses Connect.

  2. Paste the MCP server URL, leave Advanced settings empty and choose Add.

    https://schedultary.com/mcp

  3. Choose Connect. Sign in with Schedultary, pick your workplace and the access level (Admin or Read Only), then choose Allow. Staff can also check Can File My Requests and Swaps. The app then shows in Settings > Connections.

Checked September 2026. Provider Docs

Claude Code

Sign In with Schedultary

Connects over MCP. Sign in with OAuth or send your key in a header.

  1. Run this in your terminal.

    claude mcp add --transport http schedultary https://schedultary.com/mcp

  2. Start Claude Code, type /mcp, choose schedultary and then Authenticate. Sign in with Schedultary, pick your workplace and the access level (Admin or Read Only), then choose Allow. Staff can also check Can File My Requests and Swaps. The app then shows in Settings > Connections.

  3. Or use your key instead of signing in.

    claude mcp add --transport http schedultary https://schedultary.com/mcp --header "Authorization: Bearer sk_live_YOUR_KEY"

  4. Or share it with a project: save this as .mcp.json in the project folder. Replace the key with ${SCHEDULTARY_API_KEY} to read it from your environment instead.

    {
  "mcpServers": {
    "schedultary": {
      "type": "http",
      "url": "https://schedultary.com/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_YOUR_KEY"
      }
    }
  }
}

Checked September 2026. Provider Docs

Claude API

Works with a Key

The Messages API MCP connector (beta) calls Schedultary for you.

  1. Install the SDK with pip install anthropic, set ANTHROPIC_API_KEY, then run this.

    import anthropic

client = anthropic.Anthropic()
response = client.beta.messages.create(
    model="claude-opus-5-5",
    max_tokens=1000,
    messages=[{"role": "user", "content": "Who am I in Schedultary and what is on my schedule this month?"}],
    mcp_servers=[{"type": "url", "url": "https://schedultary.com/mcp", "name": "schedultary", "authorization_token": "sk_live_YOUR_KEY"}],
    tools=[{"type": "mcp_toolset", "mcp_server_name": "schedultary"}],
    betas=["mcp-client-2025-11-20"],
)
print("".join(block.text for block in response.content if block.type == "text"))

Checked September 2026. Provider Docs

Claude Code Command

claude mcp add --transport http schedultary https://schedultary.com/mcp --header "Authorization: Bearer sk_live_YOUR_KEY"

MCP Client Config

For clients that take a URL plus headers.

{
  "mcpServers": {
    "schedultary": {
      "type": "http",
      "url": "https://schedultary.com/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_YOUR_KEY"
      }
    }
  }
}

HTTP API

The same data without MCP. Every route is listed at /api/v1 and /api/v1/openapi.

curl -s https://schedultary.com/api/v1/me \
  -H "Authorization: Bearer sk_live_YOUR_KEY"

curl -s -X POST https://schedultary.com/api/v1/call \
  -H "Authorization: Bearer sk_live_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tool": "get_org", "input": {}}'

Paste This Prompt Into Your Assistant

I use Schedultary (https://schedultary.com) to build and run my team's work schedules. Help me with it through the Schedultary MCP server.

Connection
- MCP server URL: https://schedultary.com/mcp (Streamable HTTP)
- Sign in with OAuth (Sign in with Schedultary) when this app offers it. I pick my workplace and access level on schedultary.com.
- If this app needs a header instead, the connector sends Authorization: Bearer sk_live_YOUR_KEY. I create the key in Schedultary under Settings > API Keys and paste it into the connector settings myself. Never ask me to paste a key into this chat.
- If you cannot see any Schedultary tools yet, tell me step by step how to add this MCP server as a custom connector in the app we are using, then wait for me.

How to work
1. Call whoami first so you know my workplace, my role and whether this connection can make changes.
2. Read before you write. Get people, rules and requests from the list and get tools, such as get_roster, list_rules, list_requests (get_program for a rotation program, such as a residency), schedule_summary and issues_summary.
3. Apply my rules. Tell me in plain words which rules shape the schedule. Ask me before you add or change a rule.
4. Draft. When I say go, generate a draft (run_schedule_builder for a workplace, generate_year for a rotation program). Then list the Must Fix and Worth a Look issues and offer the top fix from suggest_fixes for each one.
5. Requests. List the requests still waiting and suggest a decision for each with a one-line reason. Call decide_request only after I agree.

Before you change anything
- Changes save right away and show in Schedultary's Activity. Before any change, approval, deletion or publish, tell me exactly what you plan to do and wait for my yes.
- When a tool takes dryRun, run it with dryRun: true first and show me what would change.
- Never publish a schedule unless I say publish.
- Answer me in plain words, not JSON.
Show Technical Details
  • A connected app or key acts as the person who connected it, in one workplace.
  • Staff tools include file_my_request, withdraw_my_request, propose_my_swap, offer_my_shift, pick_up_open_shift and answer_swap_offered_to_me. Limits per person: 10 requests and 5 proposals a day, 3 open proposals and 30 changes a week by default, set by the workplace, plus 20 changes in 10 minutes per assistant. Activity marks each one as filed by an assistant.
  • Supported schedule changes can be checked first. With dryRun the assistant sees exactly what would change and which Must Fix and Worth a Look issues it adds or clears, then apply_plan makes that change once if nobody changed the workplace since. simulate compares options without saving. apply_batch saves up to 25 changes together or none. An idempotencyKey makes a retry safe. Tools identify whether they support a preview.
  • check_connection reads the connected person's schedule, or the team summary for Admin connections, before confirming setup. It reports the workplace, permissions and actual schedule result. Admin summaries can include drafts. A failed read is never reported as a successful check. The Swap My Shift prompt is available to staff connections allowed to file workplace swaps.
  • Each connected app or key can make 600 calls every 10 minutes. Disconnect an app in Settings > Connections or revoke a key in Settings > API Keys. Access stops at once.
  • The MCP server lists tools such as get_grid, list_rules, add_rule, set_cell, generate_year, decide_request and list_issues, a few resources (this year's grid, rules and open issues) and prompts such as Plan Next Year. Start with whoami.
  • To understand a schedule, agents call explain_cell (why someone holds a shift or rotation), explain_issue (the rule, the numbers and why it matters), why_not (the blocking reasons in order and whether an admin can override) and suggest_fixes (up to five ranked fixes, each checked against the rules without saving). Each answers with one plain sentence first.
  • Compact reads keep answers small: schedule_summary, issues_summary and person_summary, plus fields, limit and cursor on the big lists and an ETag on every read so an unchanged answer costs nothing.
  • Scripts that cannot use MCP call the HTTP API: the catalog at /api/v1, the OpenAPI document at /openapi.json, agent flows at /agents.json and a plain summary for models at /llms.txt.

OpenAPIAgent FlowsModel Summary

HomeGuidesSupport