Manage Pipelines with MCP

peach's model context protocol (mcp) server lets a compatible ai assistant inspect and manage your pipelines using natural language requests you can ask the assistant to create a pipeline, inspect its stages, enroll a contact, move an item, configure drop off thresholds, or attach automations to stage events the assistant translates your request into peach mcp tool calls before you begin you need a peach account with pipelines enabled permission to access the relevant account and pipelines an mcp compatible client, such as chatgpt, claude, claude code, antigravity, or another remote http mcp client peach connected and authorized in that client connect peach to your ai assistant in peach, open the connect peach to your ai assistant setup page choose your ai client copy the displayed peach mcp server url add it as a custom mcp server or connector in your ai client complete the peach authorization prompt in the browser window opened by your client return to the client and ask it to list your peach pipelines the exact connection screen varies by client client where to add peach chatgpt settings → connectors → add custom connector claude settings → connectors → add custom connector claude code add a remote http mcp server with claude mcp add transport http antigravity settings → customizations → add mcp other clients add the displayed url as a remote http mcp server with oauth your client should handle peach sign in and oauth authorization do not paste passwords or access tokens into prompts for more details, checkout https //trypeach ai/mcp https //trypeach ai/mcp start by asking the assistant to orient itself pipeline operations use ids such as pipe …, stg …, and pli … you normally do not need to type them yourself ask the assistant to find the relevant pipeline, stage, and item before changing anything a good opening prompt is list my peach pipelines for each one, show its friendly item name and stages do not make any changes for changes affecting several records, ask for a preview first find all active leads in the contacted stage of the sales pipeline show me what would move to demo scheduled, but do not move anything yet after reviewing the result move the items from that preview to demo scheduled report each item that changed and any failures this two step pattern reduces mistakes when names are similar or a request affects multiple contacts what you can manage through mcp task available through mcp list pipelines yes view a pipeline and its stages yes create or update a pipeline yes create, rename, recolor, reorder, or delete stages yes list active items, optionally filtered by stage yes enroll a contact yes move an item to another stage yes update an item's title or assignee yes configure stage drop off thresholds yes attach peach automations to stage entry, exit, or drop off yes read pipeline analytics reports not currently exposed as pipeline mcp tools set item priority, next action, due date, or a general work status not supported by the current pipeline model or mcp tools common prompts discover pipelines and stages list all active pipelines in peach include each pipeline's id, friendly item label, and lifecycle settings show me the stages in the customer support pipeline in their current order explain which ones are open, won, and lost list up to 50 active tickets in the waiting for customer stage include the contact, assignee, days in stage, and drop off indicator create a pipeline create a pipeline called partner onboarding call its items partner and partners use partner as the item id prefix then add these stages in order new as open, verification as open, training as open, live as won, and cancelled as lost show me the final structure before doing anything else a newly created pipeline is empty creating stages is a separate set of operations, so ask the assistant to return the completed structure and report any stage it could not create update pipeline details rename the sales leads pipeline to enterprise sales change the friendly item labels to opportunity and opportunities do not change its stages enable re enrollment for the renewals pipeline so a customer can start a new journey after reaching won, lost, or dropped off explain what will happen to the earlier item configure the collections pipeline so a dropped off item moves to its lost stage enroll a contact add priya shah, phone number +91 90000 00000, to the sales pipeline use the first open stage do not send a whatsapp message enroll ahmed hassan, phone number +971 50 000 0000, into the qualified stage of enterprise sales and assign the item to user usr … confirm the pipeline and stage before enrolling if the phone number already belongs to a peach contact, peach uses that contact otherwise, peach creates one enrollment through mcp does not send a message only one active item is allowed for a contact in the same pipeline if re enrollment is enabled and the earlier item is terminal or dropped off, peach archives the earlier journey and creates a new one find and move an item find the active sales pipeline item for acme ltd show its current stage and readable item id do not change it move acme ltd from contacted to demo scheduled verify the destination stage belongs to the same pipeline and report the resulting stage show all active items in demo scheduled that have been there for at least seven days do not modify them the assistant should list items first when a name could match more than one contact peach mcp moves items using the item's pli … id and destination stg … id rather than guessing from names update an item's title or assignee find the active item for northwind traders in enterprise sales and change its title to “northwind — annual renewal” assign item pli … to user usr … show the updated assignee the current mcp tool accepts only a title and assignee for item updates supply a peach user id when assigning; do not ask the assistant to invent one manage stages add a stage named security review after demo scheduled in enterprise sales make it an open stage with color #6366f1 rename the stage “closed won” to “contract signed” and keep its type as won reorder the support stages to new, in progress, waiting for customer, resolved, closed first show me the current and proposed order, then wait for confirmation delete the deprecated stage from the support pipeline first confirm that it contains no active items deleting a stage fails while it still contains items or is otherwise referenced move its items and remove relevant rules before trying again configure at risk and dropped off behavior in the negotiation stage, flag leads at risk after 48 hours without activity and mark them dropped off after 168 hours without activity measure inactivity from the last activity and record “no progress in negotiation” as the drop off reason in waiting for customer, measure inactivity from our last outbound message flag tickets at risk after 72 hours and dropped off after 168 hours supported inactivity clocks are last activity at entered stage at last outbound message at drop off evaluation runs approximately every 15 minutes configuring a threshold does not immediately recalculate every item at the moment the mcp call finishes attach automations to stage lifecycle events list my peach automations and find the one called send demo preparation message attach it to the demo scheduled stage so it runs when a lead enters that stage do not replace any existing stage automations without showing them first attach automation strm … to the lost stage's drop off event preserve the stage's existing entry and exit automations stage lifecycle configuration accepts automation or stream ids beginning with strm … the mcp update replaces any lifecycle list that it supplies, so the assistant should read the current stage configuration and include existing ids that must be preserved for event driven behavior such as “enroll when a message arrives” or “move when a template is delivered,” ask the assistant to create or update a peach automation rather than expecting the stage tools to listen for those events themselves recommended prompts for recurring work morning pipeline review review the sales pipeline without making changes group active leads by stage, identify unassigned items, and highlight items marked at risk or dropped off include days in stage and readable item ids stale stage review list active items in each open stage of customer support highlight anything that has spent seven or more days in its current stage do not move or message anyone safe bulk movement find leads in qualified that match \[your criteria] return a numbered preview with contact, item id, current stage, and proposed destination wait for my confirmation before moving any item configuration audit audit the sales pipeline configuration without changing it report stage order and type, re enrollment settings, drop off thresholds, inactivity basis, and stage lifecycle automations flag missing won or lost stages prompt writing tips name the pipeline and stage clearly include a phone number when enrolling a contact say do not make changes when you want a report or preview ask the assistant to wait for confirmation before bulk moves, deletion, or reordering ask for item, pipeline, and stage ids in the result when you may continue the task later state whether existing lifecycle automations must be preserved never ask the assistant to guess an id when two records have similar names understand ids returned by mcp object id format example purpose pipeline pipe … select the pipeline to inspect or change stage stg … select a starting or destination stage pipeline item pli … move or update one contact's journey readable item id such as acm sales 42 refer to an item in conversations with your team user usr … assign an item to a peach user automation / stream strm … attach an automation to a stage lifecycle event always use ids returned by peach ids belong to one account, and stage ids belong to one pipeline troubleshooting the pipeline tools do not appear confirm that pipelines is enabled for the peach account, reconnect the peach mcp server, and ask the client to refresh its tool list peach says pipeline or stage not found ask the assistant to list pipelines again, then retrieve the selected pipeline and its stages the pipeline may have been archived, the stage may have been removed, or an id may belong to another pipeline enrollment returned an existing item the contact already has an active journey in that pipeline ask the assistant to show the existing item a duplicate active item is not created a stage cannot be deleted it still contains items or is referenced by configuration list the stage's items, move them deliberately, review connected rules, and try again an item does not appear in a list the list tool returns active items and accepts a maximum of 100 results per request narrow the request by stage, or ask for the most recently updated items archived prior journeys are not included the assistant cannot show pipeline analytics the current pipeline mcp tools manage pipeline structure and active items; they do not expose the pipeline analytics reports open the pipeline's analytics tab in peach for overview, funnel, response times, productivity, dispositions, conversions, and velocity current pipeline mcp tool reference mcp tool purpose peach list pipelines list active pipelines peach get pipeline retrieve one pipeline and its ordered stages peach create pipeline create an empty pipeline peach update pipeline update details and supported lifecycle settings peach list pipeline items list active items, optionally filtered by stage or automation status peach enroll contact in pipeline find or create a contact and enroll it without sending a message peach move pipeline item move an active item to another stage peach update pipeline item update an item's title or assignee peach create pipeline stage add a stage peach update pipeline stage rename, recolor, or change a stage type peach reorder pipeline stages set the complete stage order peach delete pipeline stage delete an unused stage peach configure stage dropoff update at risk and drop off settings peach configure stage playbooks attach automations to stage entry, exit, or drop off