Queue
Queue
Queue keeps a team's approved support response times by customer plan: which reply window, which channel, and which staffing promise may be stated, then lets an assistant read that set before it replies.
An assistant can store and look up what the team has approved for each plan. It must refuse a faster reply, a dedicated agent, or a channel that plan does not include. It does not invent a softer promise.
It works with ChatGPT, Claude, Gemini, Grok, and Cursor, plus any other MCP client that can do Streamable HTTP and OAuth. It is not a ChatGPT-only plugin.
Sign in with your Queue account when the assistant opens OAuth. Do not paste an API key or password into a header. Response tools need Pro or an active trial. The trial is 14 days, then Pro. Checkout shows the plan terms. This page does not print a price.
What the assistant can do
After you approve the connection, the server exposes these tools:
list_response_catalogs
create_response_catalog
save_customer_plan
search_customer_plans
save_reply_window
list_reply_windows
save_plan_channel
list_plan_channels
save_staffing_promise
list_staffing_promises
evaluate_support_promise
evaluate_plan_coverage
get_queue_context
evaluate_support_promise returns APPROVED only for a reply window, a channel, or a staffing promise already stored on that customer plan. Otherwise it returns REFUSED and the assistant must not promise a faster reply, a dedicated agent, or a channel the plan does not include. evaluate_plan_coverage refuses a channel or staffing promise that is outside the saved plan. The assistant only calls these tools when you and the host allow it.
Related MCP server: Review
Connect
Run the server and use its /mcp path. With the default local base, that is http://localhost:3000/mcp. A deployed host uses the same path on APP_BASE_URL.
Cursor, in ~/.cursor/mcp.json or a project .cursor/mcp.json:
{
"mcpServers": {
"queue": {
"url": "http://localhost:3000/mcp"
}
}
}Claude Code:
claude mcp add --transport http queue http://localhost:3000/mcpDo not pass an Authorization header. Other clients add the same URL, choose OAuth, and leave client id and secret empty. Queue supports dynamic client registration. Full steps for ChatGPT, Claude, Gemini, Grok, and Cursor are on the server's /connect page.
When stdin is a terminal, Queue serves Streamable HTTP. When stdin is not a terminal, it also speaks MCP on stdio. That is the same choice status-mcp makes, so a launcher can attach stdin.
Registry metadata for this server is in server.json (io.github.LAHutchins91/queue).
Run
npm ci
npm test
npm run build
npm startHosted accounts use Supabase for sign-in and response storage, and Stripe for the 14-day trial and Pro. Copy .env.example to .env and set the variables there. supabase/schema.sql creates the tables. Stripe price ids belong in the environment. Queue never displays the amount.
Local stdio, with no account configured, stores one operator's response catalog under QUEUE_DATA_DIR (or the system temp directory). HTTP tool calls still require OAuth and an active trial or Pro.
docker build -t queue-mcp .
docker run --rm -p 3000:3000 queue-mcpA container without a terminal on stdin speaks MCP on stdio and still listens on port 3000.
Available Tools
13 toolscreate_response_catalogcreate response catalogBIdempotent
Create a response catalog when the user asks for a new home for approved support response times. Does not approve a reply window, a channel, or a staffing promise.
Inputs: name (required), description. Returns: data
evaluate_plan_coverageevaluate plan coverageBRead-onlyIdempotent
Decide whether this customer plan includes a reply window, a channel, or a staffing promise. REFUSED means do not offer it. OFFER_CHANNEL refuses a channel the plan does not include. OFFER_STAFFING refuses a dedicated agent the plan does not include. Permission to state wording still requires evaluate_support_promise for the exact text.
Inputs: action (required), planId (required), catalogId (required), channel, +1 more. Returns: data
evaluate_support_promiseevaluate support promiseBRead-onlyIdempotent
Decide whether a proposed reply window, channel, or staffing promise is in the approved set for this customer plan. REFUSED means do not say it. Do not promise a faster reply, a dedicated agent, or a channel the plan does not include, and do not invent a substitute. APPROVED means sayOnly is the only permitted wording.
Inputs: kind (required), planId (required), proposal (required), catalogId (required). Returns: data
get_queue_contextget queue contextARead-onlyIdempotent
Retrieve customer plans that match a support question, plus each plan's approved reply windows, channels, and staffing promises. This is evidence, not permission. A missing plan is not an invitation to invent a faster reply, a dedicated agent, or a channel. Call evaluate_support_promise before stating any of them.
Inputs: question (required), catalogId (required), limit. Returns: data
list_plan_channelslist plan channelsBRead-onlyIdempotent
Read the channels included on one customer plan. A channel that is not listed is not included. Do not offer it.
Inputs: planId (required), catalogId (required), includeRetired. Returns: data
list_reply_windowslist reply windowsBRead-onlyIdempotent
Read reply windows for one customer plan. Listing them does not approve a faster reply. Call evaluate_support_promise with the exact wording before saying one.
Inputs: planId (required), catalogId (required), includeRetired. Returns: data
list_response_catalogslist response catalogsBRead-onlyIdempotent
List this team's response catalogs. Use the returned id. Do not guess a catalog.
Inputs: offset. Returns: data
list_staffing_promiseslist staffing promisesBRead-onlyIdempotent
Read staffing promises for one customer plan. A dedicated agent that is not listed must not be promised.
Inputs: planId (required), catalogId (required), includeRetired. Returns: data
save_customer_plansave customer planAIdempotent
Store or revise a customer plan the user has explicitly approved. The plan is the set of reply windows, channels, and staffing promises that may later be stated. Identical retries keep the same revision. A changed plan requires expectedRevision from a previous read.
Inputs: name (required), summary (required), catalogId (required), status, +1 more. Returns: data
save_plan_channelsave plan channelBDestructiveIdempotent
Store a support channel the user has explicitly approved for one customer plan. channel is email, chat, phone, sms, or video. The statement is the only channel wording that may later be approved. A channel not saved here is not included.
Inputs: name (required), planId (required), channel (required), catalogId (required), +3 more. Returns: data
save_reply_windowsave reply windowADestructiveIdempotent
Store the reply window the user has explicitly approved for one customer plan. withinMinutes is the approved wait. The statement is the only reply-time wording that may later be approved. A statement that promises a shorter wait than withinMinutes is rejected. Do not save a faster reply the user did not approve.
Inputs: name (required), planId (required), catalogId (required), statement (required), +3 more. Returns: data
save_staffing_promisesave staffing promiseBDestructiveIdempotent
Store a staffing promise the user has explicitly approved for one customer plan. SHARED_QUEUE is a shared queue. DEDICATED_AGENT is a dedicated agent. The statement is the only staffing wording that may later be approved. Do not save a dedicated agent the user did not approve.
Inputs: name (required), planId (required), catalogId (required), statement (required), +3 more. Returns: data
search_customer_planssearch customer plansARead-onlyIdempotent
Look up customer plans by name or summary. An empty result means there is no approved plan. Do not invent a reply window, a channel, or a staffing promise. Page with offset.
Inputs: catalogId (required), query, offset, includeRetired. Returns: data
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 13 tools
Most tools target distinct resources or actions, and the descriptions explicitly separate the risky pairs: evaluate_support_promise checks exact wording while evaluate_plan_coverage checks inclusion. However, search_customer_plans versus get_queue_context and the various save/list pairs still have adjacent purposes that could cause occasional misselection.
All names use snake_case verb_noun patterns, which is highly predictable. The only minor inconsistency is semantic: save_customer_plan and create_response_catalog both create/store primary records, and the set mixes search/list/get for retrieval.
With 13 tools, the server is well scoped for managing plans, reply windows, channels, staffing promises, catalogs, and evaluation checks. Each tool corresponds to a meaningful operation rather than redundant granularity.
The surface covers creation and reading of plans, reply windows, channels, and staffing promises, plus important approval checks. But it lacks delete/revoke operations for approved promises or channels, and response catalogs have only list/create with no update or delete, which may leave agents unable to retract or clean up records.
Maintenance
Related MCP Connectors
Verified AI free tiers plus source-backed agent and MCP discovery monitoring.
15 tools, 38 languages, 8 AIs. Apple-approved 3-day Pro trial; subscription renews, no free Sends.
Turn your offers into an AI agent. Set it up from your MCP assistant, answer customer questions, collect enquiries and connect with other agents via A2A. OAuth sign-in; free plan available.
Post and schedule to TikTok, Instagram, YouTube, Facebook, LinkedIn, X, Threads and Bluesky from your AI agent. Video included, you approve every post, and MCP is on the free plan.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceClaim by Ouroboros Apps (https://ouroborosapps.com). Brand claims and marketing copy compliance: approved claims, offers, proof, brand voice, and banned phrases, with a check that rejects copy inventing a guarantee or discount. Connect by pasting https://claim-continuity2.vercel.app/mcp into Claude, Cursor, Grok, or ChatGPT developer mode (OAuth). Docs: https://ouroborosapps.com/docs/claim53 npmMIT
- AlicenseAqualityFmaintenanceApproved public replies to customer reviews. An assistant can only send a reply that matches stored wording. A 14-day trial, then Pro. The price is only at checkout.5MIT
- AlicenseAqualityFmaintenanceApproved product specs, prices, and availability for shopping assistants over MCP. A 14-day trial, then Pro. The price is only at checkout.5MIT
- AlicenseAqualityFmaintenanceApproved curriculum for coaches, connected to AI assistants over MCP. A 14-day trial, then Pro. The price is only at checkout.12MIT