MCP App Store
by KeenEthics
Kanal icon

Kanal MCP Server for ChatGPT

By Kanal

Overview

Kanal is a WhatsApp marketing platform for e-commerce brands. Connect your Kanal account to ask ChatGPT anything about your WhatsApp marketing performance: attributed revenue, ROAS and total channel cost, campaign and automation stats down to individual flow steps, audience segments and sizes, and Meta template approval status. All tools are read-only, and every metric uses the same attribution engine and settings as your Kanal dashboard, so the numbers always match what you see in the app. Example questions: - "What's my WhatsApp ROAS over the last 30 days?" - "Compare my last 3 campaigns" - "Which step of my abandoned-cart flow converts best?". Requires a Kanal account (Scale plan for full access, discovery quota on Pro).

How to connect

  1. ChatGPT: Open the Kanal listing in the ChatGPT app directory and choose Connect. ChatGPT listing Kanal exposes 64 tools here.

Tools

add_contacts_to_group

ChatGPT
Adds contacts to a static group. A group is channel-agnostic: WhatsApp and Instagram contacts can share it (list_segments_and_groups only counts the WhatsApp-reachable ones). Idempotent: contacts already in the group are reported as already_in_group, ids that do not belong to the team as unknown_contact_ids. Returns the number actually added.

add_nodes

ChatGPT
Adds one or more nodes to a flow in a single transaction (an active automation needs confirm_active). Each node: type + data (fields per type: get_automation_capabilities) and optionally after = the node to chain from (an existing node_id, or the $ref of a previous item of this same call). after connects from that node's default handle (its main path); use after_handle to chain from a branch (conditions: success/failure, ab-testing: test_a/test_b, wa-message: a button handle — $ref_button_1 for a node of this same call, {node_id}_button_1 otherwise —, wa-input: waiting_input). Positions follow the builder layout (left to right, branches stacked to the right of their node, next step under them); omit them. Nothing is saved if any node is invalid. Returns the new node ids with their concrete handles; use them with connect_nodes for extra branches. Max 25 nodes per call.

confirm_automation_activation

ChatGPT
Activates the automation described by a prepare_automation_activation token: from then on it runs for every new trigger event and messages real customers. Call it only after the user explicitly confirmed the summary, never on your own initiative. Refused if the token expired (10 min), was already used (a refused call uses it too), or if the flow changed since. The result lists the other active automations on the same trigger.

confirm_campaign_send

ChatGPT
Schedules (or starts) the campaign described by a prepare_campaign_send token. This spends real money and messages real customers: call it only after the user explicitly confirmed the summary, never on your own initiative. Refused if the token expired (10 min), or if the flow or the audience changed since. Once scheduled, changes happen in Kanal (reschedule, pause).

connect_nodes

ChatGPT
Creates an edge from an output handle of source_node_id to target_node_id. Omit source_handle for the node's main path. Branch handles: conditions success/failure, ab-testing test_a/test_b, wa-message a button handle from add_nodes, wa-input / wa-ask-list-option waiting_input (loop back while waiting), voucher-generate / api-call failure. Re-sending an identical connection returns created=false.

create_automation

ChatGPT
Creates an INACTIVE automation draft with its trigger node pre-configured; then add_nodes with after = trigger_node_id. Trigger types and their trigger_config: see get_automation_capabilities (welcome/keyword flows → kanal_v2_keyword with keywords; post-purchase → shopify_order_created; abandoned cart → shopify_abandoned_checkout; shipping → shopify_shipment_updated). Several automations may share a trigger; the result lists the active ones already on it so you can tell the user. Abandoned-checkout flows stop by themselves as soon as the customer orders: never add a 'has ordered' condition. Fails if an automation with the same name exists. Activation is a separate confirmed step (prepare_automation_activation).

create_campaign

ChatGPT
Creates a flow campaign DRAFT (step CREATED, no audience, no schedule) and returns its automation_id and trigger_node_id. Compose the messages with add_nodes on that campaign_id: the first node after the trigger must be a wa-template-v2 (or wa-template-carousel) using a template written for this campaign (create_template + submit_template, PENDING is fine) or one the user names; session messages may follow a template button. Audience, scheduling and sending are done by the user in Kanal at builder_url. Fails if a draft campaign with the same name exists.

create_group

ChatGPT
Creates an empty static group (a manually managed contact list, usable as a campaign audience). Fill it with add_contacts_to_group. Fails if a group with the same name already exists. Optional color as a hex code (default #3B82F6).

create_media_upload

ChatGPT
Use when the media the user wants in a template or a flow has no public URL (an image they attached in the chat, a file on their computer). Returns an upload_url valid 30 minutes for up to 10 files (JPG/PNG ≤ 5 MB, MP4 ≤ 16 MB). If you can run code with network access and hold the file, POST it yourself as multipart/form-data, field "file" (curl -F "file=@photo.jpg" <upload_url>); the response gives its url. Otherwise give the upload_url to the user: opened in a browser it is a page where they drop the files. Then call get_media_upload for the Kanal URLs of what was received, and pass them as header_media_url (create_template, update_template_header_media), card media_url, or a wa-image / wa-video node url.

create_segment

ChatGPT
Creates a dynamic audience segment from conditions (format: see get_segment_capabilities; check the count first with preview_segment). The segment is calculated asynchronously right after creation and then kept up to date; use list_segments_and_groups to read its contact count. Fails if a segment with the same name already exists.

create_template

ChatGPT
Creates a WhatsApp template as a DRAFT in Kanal (nothing is sent to Meta). Body placeholders are positional ({{1}}, {{2}}…) and map in order onto variables, whose names are resolved at send time: recipient fields (name, first_name, last_name, phone) or trigger / campaign payload keys (order_number, tracking_url…; see get_available_variables). Then submit_template sends it to Meta for approval (minutes to hours) — ask the user before submitting. A campaign can reference a PENDING template; it is held until approval. Write a dedicated template for each template message of a flow, with the real copy the user wants; reuse one from list_templates only when the user names it or its body (read with get_template) already matches. An existing template keeps its text: only its header media and variable mapping can change in place (update_template_header_media, update_template_variables); anything else means a new template under a new name. Pass automation_id or campaign_id when the template is written for a flow (the usual case): it then belongs to that flow like a template created from the builder, instead of showing up as a standalone template in Settings. A message the user wants sent to customers (launch, promotion, news) is a campaign: create_campaign first, then this tool with its campaign_id; a template alone is not something the user can send. The header is either text (header_text) or a public JPG/PNG/MP4 URL (header_media_url, e.g. a Shopify product image_url from search_products); the file is copied into Kanal. A carousel (several products side by side) is carousel_cards: 2 to 10 cards under the shared body_text, each with its own image or video, a short text and the same buttons; send it with a wa-template-carousel node.

deactivate_automation

ChatGPT
Turns an automation OFF: new trigger events are ignored from then on. Side effect to tell the user before calling: a wait that expires or a reply received while the flow is off ends the contact's session; the others resume on reactivation. The result reports contacts_waiting = {total, waiting_for_delay: {total, by_node: {wait node id: contacts}}, waiting_for_reply: contacts asked a question (wa-input) who have not answered yet}; contacts sent buttons they have not tapped are not counted. Idempotent: an inactive automation is left as it is.

get_account

ChatGPT
Returns the connected Kanal team: name, subscription plan, the messaging channels it has connected — WhatsApp phone numbers (with Meta quality rating) and Instagram accounts (username, connected/disconnected) — and the attribution settings (basis, window, timezone) that all metric tools use. A phone number with whatsapp_connected=false is an empty channel line created at signup, not a broken number: the team never finished WhatsApp onboarding, and the line may still carry an Instagram account. Call this first to understand the account context, in particular which channels exist before reading the inbox.

get_automation_capabilities

ChatGPT
Describes everything the flow tools accept: trigger types (with their trigger_config), node types available through MCP with their fields, default output handle and branch handles, the conditions DSL, the variables syntax, the campaign rules (a campaign must send an approved template before any session message) and worked examples. Call it once before create_automation / create_campaign / add_nodes. Pass node_type for one type in full.

get_automation_graph

ChatGPT
Returns an automation (or the flow behind a campaign): name, trigger, status, graph_version, every node with its data, label, output handles and outgoing edges, and the builder_url. Use it before update_node / connect_nodes and after any error to see the current state.

get_automation_stats

ChatGPT
Money + engagement stats for a single automation flow (attributed revenue, cost, cost_marketing, ROAS, orders, delivered/read/clicked) over a period, plus a per-node breakdown, using the team attribution settings. cost = spend across ALL WhatsApp conversation categories, marketing AND utility (the ROAS denominator — marketing is routinely routed as utility for Meta pricing); cost_marketing = the marketing-only slice of that spend. Find automation ids with list_automations.

get_available_variables

ChatGPT
Returns the {{...}} tokens a flow can render: contact fields ({{recipient.name}}…), trigger payload fields mapped by Kanal ({{variables.shopify_abandoned_checkout_url}}…), and values produced upstream (wa-input answers, vouchers, api-call mappings) when after_node_id is given. Any other token is refused when the node is written. Call it before writing a message that uses variables.

get_best_practices

ChatGPT
Recommendations written by the Kanal team: how to write WhatsApp templates that get approved and read, campaign timing and audience rules, flow patterns that convert, opt-in and compliance. Read them before designing a campaign, a flow or a template, and follow them unless the user asks otherwise. Categories: whatsapp, templates, campaigns, automations, audience, kanal.

get_billing

ChatGPT
Returns what the team pays Kanal for: the subscription (plan, provider), the WhatsApp conversation billing (prepaid balance in euros, auto-refill amount and threshold, conversations billed and top-ups of the last 30 days, the free service-message allowance) and, for teams with the AI assistant, the AI credits of the current month (included, consumed, remaining, overage and its cost, hard cap; absent otherwise). Answers "how many credits do I have left", "what did the AI cost me this month", "what is my auto-refill". Read-only: changing the auto-refill, applying a promo code or opening the payment portal is done by the user at billing_page_url (give them the link, never fetch it). No card or Stripe identifiers are returned.

get_campaign_cost_estimate

ChatGPT
Estimates what sending a flow campaign to its current audience would cost in Meta conversation fees (EUR): cheapest and most expensive path through the flow, per-contact maximum, and the templates involved with their approval status. Zero until set_campaign_audience has been called. Read-only.

get_campaign_stats

ChatGPT
Money + engagement stats for a single campaign (attributed revenue, cost, cost_marketing, ROAS, orders, delivered/read/clicked) over a period, using the team attribution settings. cost = spend across ALL WhatsApp conversation categories, marketing AND utility (the ROAS denominator — marketing is routinely routed as utility for Meta pricing); cost_marketing = the marketing-only slice of that spend. Find campaign ids with list_campaigns.

get_contact_import_status

ChatGPT
Progress and report of an import_contacts run: status (pending, importing, completed, failed), rows imported, skipped (invalid numbers) and duplicates. Poll every few seconds until completed.

get_conversation

ChatGPT
Returns the most recent messages of one conversation in chronological order (oldest first), with a contact header (name, phone number and/or Instagram username shown in full, channels, opt-in status). When the contact talks on both channels the two threads are merged into one timeline; each message carries its channel. Each message has: direction (in/out), type, a text body (media and interactive replies collapse to a bracketed placeholder like "[image]" or "[button: Yes]", a deleted Instagram message to "[deleted message]"), who sent it (source: on WhatsApp human = a team member from the inbox, ai = the Kanal AI agent, automation = a flow (with automation_id), campaign = a campaign blast or its follow-ups (with campaign_id), integration = a helpdesk relay such as Gorgias or this connector (via: gorgias, crisp, mcp…); on Instagram, kanal or instagram_app for messages the team sent from Meta's own apps), an Instagram reaction emoji when there is one, and the timestamp. To answer, use send_message. Find contact ids with list_conversations. Args: contact_id (required), channel (all, whatsapp, instagram; default all), limit (default 50, max 100).

get_media_upload

ChatGPT
Lists the files received on a link from create_media_upload, in upload order, with the Kanal url to pass as header_media_url, card media_url or a wa-image / wa-video node url. Empty files means nothing was uploaded yet: ask the user whether they dropped the files on the page.

get_performance_chart

ChatGPT
Daily time series of one metric (revenue, orders, conversations, cost, delivered, read, roas, conversion_rate) over a period, using the team attribution settings. Ideal to analyse trends or build charts. For weekly or monthly trends, aggregate the daily points (or use query_stats with target=attribution.* and group_by=week|month).

get_performance_overview

ChatGPT
Global WhatsApp marketing performance for the team over a period: attributed revenue, cost, cost_marketing, ROAS, orders, engagement funnel (delivered/read/interacted/converted), revenue split by mechanism (campaigns vs automations vs AI), and top campaigns/automations. cost = spend across ALL WhatsApp conversation categories, marketing AND utility (the ROAS denominator — marketing is routinely routed as utility for Meta pricing); cost_marketing = the marketing-only slice of that spend. Uses the team's configured attribution window and basis (see get_account). Data may lag ~30 minutes.

get_segment_capabilities

ChatGPT
Describes the conditions format accepted by preview_segment and create_segment: condition types (contact_events, properties, group_membership), events, operators, time intervals and their options, AND/OR semantics, with examples. Call this before building your first segment. Segments target WhatsApp-reachable contacts only — an Instagram-only profile is never part of one; use a group to collect those.

get_stats_capabilities

ChatGPT
Discover what query_stats can do: available targets, metrics, filterable fields and periods. Call this before your first query_stats.

get_template

ChatGPT
Returns the full content of one template from list_templates: header (text, or the media URL), body text with its {{n}} placeholders and the variables they map to, footer, quick replies, URL buttons (and the button types the MCP cannot write), carousel cards. Use it to show a template to the user, to check whether an existing one already says what a flow needs, or to recreate it: recreatable says whether create_template can take this content as is, and unsupported_by_create_template lists what it cannot. category is the category submitted to Meta (Meta may have reclassified it) and language defaults to fr on old templates.

import_contacts

ChatGPT
Creates or updates contacts from rows (phone in international format, optional name, email and custom variables) and puts them in a group, new or existing — the MCP equivalent of a CSV import; use it when the user shares a list of numbers or a file. Up to 2000 rows per call, split larger lists. Runs in the background: poll get_contact_import_status with the returned import_id. consent=true only if the user states these people opted in to WhatsApp messages (sets them opted_in); otherwise they stay "unknown" and are still reachable. Existing contacts keep their opt-out. The group can then be a campaign audience.

list_automations

ChatGPT
Lists the team's automations (visual flows): id, name, trigger type (kanal_keyword, shopify_order_created, shopify_abandoned_checkout…), active status. Use the id with get_automation_stats. Optional is_active filter and limit (default 25, max 100).

list_campaigns

ChatGPT
Lists the team's WhatsApp campaigns, most recent first: id, name, status (CREATED, SENDING, SCHEDULED, PAUSED, SENDED), scheduled date. Use the id with get_campaign_stats. Optional status filter and limit (default 25, max 100).

list_conversations

ChatGPT
Lists the team's inbox conversations across channels, most recent first: contact id, name, the channels that conversation uses, phone number and/or Instagram username (shown in full — the merchant owns these conversations and the inbox permission was granted separately at connect time), the last message (preview, channel, direction in/out, date), an unread flag, and whether the conversation is flagged for a human. Read-only. Note: "all" is broader than the in-app inbox default — it includes closed conversations and ones the business has not replied to yet. Optional channel (all, whatsapp, instagram; default all), filter (all, unread, human_needed; default all) and limit (default 20, max 50). Unread means: manually marked unread, or the last WhatsApp message is an unanswered inbound one; on Instagram there is no inbound read receipt, so it means the last Instagram message is an inbound one (the business has not replied since). Pass a contact id to get_conversation to read the full thread.

list_segments_and_groups

ChatGPT
Lists the team's audience building blocks with contact counts: segments (dynamic, condition-based) and groups (static lists). Both can be used as a campaign audience. Counts are WhatsApp-reachable contacts, exactly like the Kanal UI: a group can also hold Instagram-only profiles, which are not counted here and are skipped when a WhatsApp campaign is sent to that group.

list_teams

ChatGPT
Lists the Kanal teams (brands, shops) this connection can act on: id, name, MCP access (full = read and write, read = read only, none = plan without MCP) and whether it is the team used when a call names none. An agency or a Kanal staff member reaches several teams: call this first, pass the team id as team_id on every other call, and say which team you are working on before any change. Kanal staff reach every team: search them with query (name or id).

list_templates

ChatGPT
Lists the team's WhatsApp message templates with their Meta approval status (DRAFT, PENDING, APPROVED, REJECTED) — a template must be APPROVED before it can be sent in a campaign. Also returns category (MARKETING, UTILITY, AUTHENTICATION) and language. Use it to check approval status or to find a template the user names, then get_template for its content (text, media, buttons); for a new flow, write the message with create_template instead of picking an unrelated approved template.

prepare_automation_activation

ChatGPT
Runs every check the activation applies (graph valid, every node configured, templates submitted to Meta, 24h window, plan, Instagram ad welcome message) and returns a summary with a confirmation_token valid 10 minutes. NOTHING IS ACTIVATED by this call. Show the summary (trigger, warnings, other active automations on the same trigger) to the user, and only call confirm_automation_activation after their explicit yes.

prepare_campaign_send

ChatGPT
Runs every pre-send check on a flow campaign (audience set, templates submitted, flow valid), prices it, and returns a summary with a confirmation_token valid 10 minutes. NOTHING IS SENT OR SCHEDULED by this call. Show the summary (audience, date, cost) to the user and only call confirm_campaign_send after they explicitly confirm, or when they click Confirm on the card. scheduled_at is "now" or an ISO 8601 date; without a UTC offset it is read in timezone (default: the connected user's timezone).

preview_segment

ChatGPT
Dry run of a segment: validates the conditions and returns how many contacts match right now, without saving anything. Use it to check a definition (and show the count to the user) before create_segment. Conditions format: see get_segment_capabilities.

query_stats

ChatGPT
Free-form stats query engine (the same one Kanal's built-in assistant uses). Call get_stats_capabilities first to discover available targets, metrics, filters and periods, then query e.g. target=attribution.global or attribution.top_campaigns. group_by day/week/month on attribution targets returns a revenue time series (data[] of {period, ...}); group_by=hour is rejected (attribution is daily) and top_* stays a ranked list. Results follow the team's attribution settings and match the Kanal dashboard.

remove_contacts_from_group

ChatGPT
Removes contacts from a static group (the contacts themselves are kept). Idempotent: contacts not in the group are reported as not_in_group, ids that do not belong to the team as unknown_contact_ids. Returns the number actually removed.

remove_edge

ChatGPT
Removes one edge by id (ids come from add_nodes, connect_nodes and get_automation_graph). The nodes stay in place.

remove_node

ChatGPT
Removes a node and every edge attached to it. Trigger nodes cannot be removed. Downstream nodes stay in the graph, disconnected — reconnect them with connect_nodes or remove them too (orphaned_node_ids lists them).

report_unsupported_request

ChatGPT
Call this when the user asks for something no Kanal tool covers (a setting, an action, a piece of data), BEFORE telling them it is not possible here. It records the request for the Kanal team and changes nothing in the account. Then answer the user honestly: say it is not available through this connector today, and point them to the Kanal app when it can be done there. Args: request (the user's words), missing (what would be needed, one sentence), category.

search_contacts

ChatGPT
Finds contacts of the team whose name, phone number or Instagram username contains the query (min 2 characters), most recently active first: contact id, name, the channels the contact is reachable on, phone number (in full, null for an Instagram-only profile), Instagram username, opt-in status. Use the ids with add_contacts_to_group, remove_contacts_from_group and set_contact_variable. Limit default 20, max 50.

search_products

ChatGPT
Finds products in the team's Shopify store(s) by title or SKU fragment, live from Shopify (no query = the first products of the catalogue): product_id, title, status (ACTIVE, DRAFT, ARCHIVED), image_url, store_domain and the variants (variant_id, title, price). Use product_id as the value of a filter_on_product_ids condition (with store_domain when the team has several stores), image_url as a template header_media_url, and the price in a message. Never guess a product id: search first. Limit default 20, max 50 per store.

send_message

ChatGPT
Sends one text message to a contact, as a reply in the inbox (WhatsApp or Instagram). Session message only: WhatsApp needs the contact to have written in the last 24h; when that window is closed and the number has an approved invitation template, the message is queued and the contact receives an invitation, the text leaves when they answer (status queued). Instagram needs the window open, nothing is queued. get_conversation on this contact is REQUIRED first (within 30 minutes, and again if the contact wrote since): the call is refused otherwise. Write in the contact's language, and only send what the user asked to send: this reaches a real customer and cannot be unsent. Args: contact_id, text (max 1000 characters), channel (whatsapp or instagram; required when the contact has both).

set_campaign_audience

ChatGPT
Targets a campaign draft at a Kanal group (static list) or segment (dynamic). Replaces any previous audience; only WhatsApp-reachable contacts are kept. Nothing is sent. Returns the contact count and the cost estimate. Find or build the audience with list_segments_and_groups, create_segment / preview_segment, create_group + add_contacts_to_group, or import_contacts. refresh_before_send recomputes a segment right before sending.

set_contact_variable

ChatGPT
Creates or updates one custom variable on a contact (text value). Variables are usable as {{recipient.NAME}} in messages and as segment/automation inputs. Name: letters, digits and underscores only. Idempotent upsert; returns whether it was created or updated.

submit_template

ChatGPT
Sends a DRAFT template to Meta for review. IRREVERSIBLE: name, category and language are locked by Meta once submitted, and the review takes minutes to hours. Always show the user the template text and get their explicit go-ahead before calling this. Returns status PENDING; list_templates shows APPROVED / REJECTED later. A rejection keeps the template editable in Kanal.

submit_upsell_template

ChatGPT
Submits to Meta the limited-time offer template of a shopify-upsell-offer node, built from the node texts (offer_text, body_text, body_variables, button_text, decline_text, header image). IRREVERSIBLE like submit_template: show the user the texts and get their explicit go-ahead first. While the texts are unchanged the current template is returned without a new submission; after a change a new template is submitted under a new name. Offers leave through not_eligible until Meta APPROVES it (list_templates, validate_automation). Scale plan.

update_attribution_settings

ChatGPT
Sets the attribution defaults of the team: basis (delivered, read or clicked: which message event opens the window in which an order is attributed to a message) and window_seconds (12h to 5 days). They are the defaults of the dashboards and of every metric tool when no basis / window is passed; the rollup is already computed for every combination, so nothing is recomputed. Omitted fields are kept. Current values: get_account.attribution.

update_automation_settings

ChatGPT
Updates the settings of an automation without touching its graph: re-entry rule (always / never / delay with re_entry_delay_value + re_entry_delay_unit), stop_on_contact_reply (the contact leaves the flow as soon as they write — the way to exit a winback or a reminder sequence once the customer answers; there is no "has replied" condition node), stop_on_agent_reply (same when an agent answers from the inbox) and, on shopify_abandoned_checkout flows only, abandoned_cart_grace_hours (0 disables). Only the keys given change; get_automation_graph returns the current values. Allowed on an active automation: nothing is sent and the graph stays as it is.

update_node

ChatGPT
Replaces the configuration (data) of one node; the full data object is required (read the current one with get_automation_graph). Trigger nodes can be updated too (e.g. change keywords). Edges are kept, except button branches whose button disappeared, reported as removed_edges.

update_optout_settings

ChatGPT
Changes the opt-out detection of the team: detection_enabled (an inbound message equal to a keyword opts the contact out and sends the confirmation of the opt_out system automation, created on first enable) and custom_keywords (the full custom list, replaces the current one, max 50 characters each; the built-in stop / unsubscribe / arreter… keywords always apply). Omitted fields are kept. Read the current state in get_account.opt_out first.

update_quiet_hours

ChatGPT
Sets the team business hours and switches the quiet-hours auto-reply on or off: a message received outside the hours gets the reply of the quiet_hours automation (once per quiet period, when no other automation or the AI handled it). It never holds campaigns or flows. Omitted days keep their current slots; a day given with an empty list is closed all day. Enabling creates the quiet_hours automation if the team has none (edit its message with the graph tools) and activates it; disabling deactivates it. Read the current state in get_account.quiet_hours first.

update_template_header_media

ChatGPT
Replaces the header media (image or video) of an existing template, APPROVED ones included, without resubmitting it to Meta: the media is sent as a link at send time, Meta never locks it. The header type cannot change (an IMAGE header takes a JPG/PNG, a VIDEO header an MP4) and a TEXT header, a GIF header or a carousel cannot be changed here. Takes effect on every flow, campaign and API call that sends this template, immediately. To change the text, the buttons or the number of variables, write a new template with create_template instead.

update_template_variables

ChatGPT
Changes which value each {{n}} placeholder of an existing template receives (the variable name: name | first_name | last_name | phone | a trigger or campaign payload key, see get_available_variables) and the example shown to Meta, APPROVED templates included, without resubmitting: Meta never sees variable names. The set of placeholders is fixed: give every existing number and no other; adding or removing one changes the text and needs a new template (create_template). Takes effect immediately on every flow and the public API that send this template; a rename is refused while an unsent classic campaign (template mode) uses it, because its contacts carry values under the old names.

update_timezone

ChatGPT
Sets the team timezone (IANA name, e.g. Europe/Paris). It drives the business hours of the quiet-hours auto-reply, the days of every metric tool and dashboard, and the scheduling of campaigns: change it only when the user asks for it or the current one is clearly wrong for their business. Current value: get_account.quiet_hours.timezone.

update_utm_tracking

ChatGPT
Changes the UTM tracking of the team: enabled (Kanal appends utm_source, utm_medium and utm_campaign to every button and CTA link it sends), utm_source and utm_medium (defaults kanal / whatsapp; an empty string goes back to the default). utm_campaign is always the campaign or flow name and cannot be set. Omitted fields are kept. Read the current state in get_account.utm_tracking first.

validate_automation

ChatGPT
Runs the checks Kanal applies before activation or sending: unreachable or dangling nodes, unconfigured nodes, invalid handles, template status, the 24h conversation-window rule, campaign rules (a campaign must send a template) and nodes locked by the plan. Returns valid=true or the issues with node ids and severity. Always call it at the end of a build and fix what it reports.

Capabilities

Writes

Example Prompts

Click any prompt to copy it.

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/kanal"><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/kanal)