Cosend
REMOTE · MCP.COSEND.APP
WhatsApp Business inbox for AI apps: read and answer chats, run campaigns, deploy automations.
Available components
Last scanned (19 hours ago)
Scores higher than 84% of remote servers
Reachable. Last reached 21 Nov 2026.
Category Communication
How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. How we score → Why this is hard to score →
Endpoint Security97
- The endpoint's TLS certificate is valid, in date, and uses a strong key. View diagnostics → Pass
- Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token. View diagnostics → Pass
- HTTPS is enforced; there's no plaintext access path. View diagnostics → Pass
- The HSTS (Strict-Transport-Security) header is present. View diagnostics → Pass
- DNSSEC check failed: this domain isn't protected by DNSSEC. See how to fix → View diagnostics → Fail
- The authorisation server supports Client ID Metadata Documents, the current MCP client-registration mechanism. View diagnostics → Pass
Transport & Reachability100
- Verified streamable-http transport via a live MCP handshake. View diagnostics → Pass
Schema Quality & AI Usability59
- AI-judged instruction clarity (good).Pass
- Context-footprint check failed: tool/resource definitions use about 30302 tokens (~242/item across 125 items; 125 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management50
- Stability check failed: schema churn in the 18 days we've observed: 2 tool removals, 1 breaking changes, 0 auth/transport breaks, 102 additions. See how to fix → Fail
Tool Coverage91
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 70% of tool parameters carry a description.Partial
- Structured output schemas are declared (1% of tools); any adoption earns full credit.Pass
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- All 6 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.Pass
- An AI judge read all 126 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
- Implements a current MCP spec version (2026-07-28).Pass
Checks we ran (13)
- TLS certificate
- Authorization
- HTTPS
- HSTS header
- DNSSEC
- Transport
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Does not apply to this hosted endpoint (2)
- Supply Chain SecurityScored on package channels only. This is a hosted endpoint, so there is nothing to install or inspect.
- Provenance & TransparencyScored on package channels only. This is a hosted endpoint, so there is nothing to install or inspect.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 21 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 46 to 50.
- 20 Nov 26 0
- Tool “rollback_automation” rewrote its description, which is the text the model reads security
- Tool “switch_on_automation” rewrote its description, which is the text the model reads security
- 19 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 40 to 43.
- 18 Nov 26 0
- Tool “get_inbox_overview” rewrote its description, which is the text the model reads security
- “update_contact” reworded the description of “attributes” cosmetic
- “upsert_contact” reworded the description of “attributes” cosmetic
- 17 Nov 26 −1
- Stability: 0.43 → fail ▼ security
- A breaking change shipped without a version bump: still 1.0.0 ▼ security
- Tool “automation_recipes” was removed ▼ security
- Tool “describe_automation” was removed ▼ security
- Tool “draft_reply” was removed ▼ security
- Tool “get_authoring_chat” was removed ▼ security
- Tool “get_automation_catalogue” was removed ▼ security
- Tool “manage_automation” was removed ▼ security
- Tool “set_automation_state” was removed ▼ security
- Tool “translate_message” was removed ▼ security
- Tool “upsert_automation” was removed ▼ security
- The server rewrote its instructions, which are the text every model session reads security
- New tool “buy_cosend_number”, which the server declares destructive security
- New tool “copy_automation”, which the server declares destructive security
- New tool “create_custom_automation”, which the server declares destructive security
- New tool “delete_automation”, which the server declares destructive security
- New tool “keep_cosend_number”, which the server declares destructive security
- New tool “rollback_automation”, which the server declares destructive security
- New tool “set_up_accurate_warehouse_arrival”, which the server declares destructive security
- New tool “set_up_ai_receptionist”, which the server declares destructive security
- New tool “set_up_appointment_reminder”, which the server declares destructive security
- New tool “set_up_lead_capture”, which the server declares destructive security
- New tool “set_up_missed_call_text_back”, which the server declares destructive security
- New tool “set_up_order_confirmation”, which the server declares destructive security
- New tool “set_up_order_updates”, which the server declares destructive security
- New tool “set_up_payment_reminder”, which the server declares destructive security
- New tool “set_up_quote_intake”, which the server declares destructive security
- New tool “set_up_review_request”, which the server declares destructive security
- New tool “set_up_welcome_menu”, which the server declares destructive security
- New tool “stop_cosend_number”, which the server declares destructive security
- New tool “switch_off_automation”, which the server declares destructive security
- New tool “switch_on_automation”, which the server declares destructive security
- New tool “update_accurate_warehouse_arrival”, which the server declares destructive security
- New tool “update_ai_receptionist”, which the server declares destructive security
- New tool “update_appointment_reminder”, which the server declares destructive security
- New tool “update_automation_details”, which the server declares destructive security
- New tool “update_custom_automation”, which the server declares destructive security
- New tool “update_lead_capture”, which the server declares destructive security
- New tool “update_missed_call_text_back”, which the server declares destructive security
- New tool “update_order_confirmation”, which the server declares destructive security
- New tool “update_order_updates”, which the server declares destructive security
- New tool “update_payment_reminder”, which the server declares destructive security
- New tool “update_quote_intake”, which the server declares destructive security
- New tool “update_review_request”, which the server declares destructive security
- New tool “update_welcome_menu”, which the server declares destructive security
- Tool “deploy_automation” rewrote its description, which is the text the model reads security
- Tool “get_report” rewrote its description, which is the text the model reads security
- Tool “import_contacts” rewrote its description, which is the text the model reads security
- Tool “manage_number” rewrote its description, which is the text the model reads security
- Tool “run_automation” rewrote its description, which is the text the model reads security
- Tool “send_campaign” rewrote its description, which is the text the model reads security
- Tool “update_campaign” rewrote its description, which is the text the model reads security
- Schema quality: 215 → 241 ▼ functional
- “deploy_automation” added a required parameter “actions”, so existing callers break ▼ functional
- “run_automation” added a required parameter “actions”, so existing callers break ▼ functional
- First check of Tool coverage: 1 functional
- New tool “list_automation_building_blocks” functional
- New tool “list_automation_recipes” functional
- New tool “list_cosend_numbers” functional
- “import_contacts” added an optional parameter “details” cosmetic
- “import_contacts” added an optional parameter “new_fields” cosmetic
- “update_campaign” added an optional parameter “ab” cosmetic
- “update_campaign” added an optional parameter “buttons” cosmetic
- “update_campaign” added an optional parameter “cards” cosmetic
- “update_campaign” added an optional parameter “header” cosmetic
- “update_campaign” added an optional parameter “kind” cosmetic
- “update_campaign” added an optional parameter “let_join” cosmetic
- “update_campaign” added an optional parameter “reply_rules” cosmetic
- “update_campaign” added an optional parameter “steps” cosmetic
- “update_campaign” added an optional parameter “track_clicks” cosmetic
- “get_report” reworded the description of “card” cosmetic
- “manage_number” reworded the description of “action” cosmetic
- 16 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 40 to 43. That category is still filling its 30-day observation window: 12 days of observed history at the previous scan, 13 at this one. The score rises as the window fills, whether or not the server changes.
- 14 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 33 to 37. That category is still filling its 30-day observation window: 10 days of observed history at the previous scan, 11 at this one. The score rises as the window fills, whether or not the server changes.
- 12 Nov 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 27 to 30. That category is still filling its 30-day observation window: 8 days of observed history at the previous scan, 9 at this one. The score rises as the window fills, whether or not the server changes.
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Probed https://mcp.cosend.app/mcp
TLS valid
Negotiated TLS 1.3 with TLS_AES_128_GCM_SHA256 .
| Subject | Issuer | Valid from | Valid until | Key | Signature | Serial |
|---|---|---|---|---|---|---|
| CN=mcp.cosend.app | CN=WE1,O=Google Trust Services,C=US | 29 Oct 2026 | 27 Jan 2027 | ECDSA 256 | ECDSA-SHA256 | e6d88c9d95117b110e3d3f2b21818bf1 |
| SANs: mcp.cosend.app | ||||||
| CN=WE1,O=Google Trust Services,C=US (CA) | CN=GTS Root R4,O=Google Trust Services LLC,C=US | 13 Dec 2023 | 20 Feb 2029 | ECDSA 256 | ECDSA-SHA384 | 7ff31977972c224a76155d13b6d685e3 |
| CN=GTS Root R4,O=Google Trust Services LLC,C=US (CA) | CN=GlobalSign Root CA,OU=Root CA,O=GlobalSign nv-sa,C=BE | 15 Nov 2023 | 28 Jan 2028 | ECDSA 384 | SHA256-RSA | 7fe530bf331343bedd821610493d8a1b |
Background: What to check on a remote MCP endpoint →
DNSSEC insecure
Validation of mcp.cosend.app. — Not signed
| Zone | DS | Keys | Algorithms | Outcome |
|---|---|---|---|---|
| . | Trust anchor | 20326, 38696 | 8, 8 | Verified |
| app. | DS published | 23684 | 8 | Verified |
| cosend.app. | No DS | Unsigned (proven) parent-signed NSEC/NSEC3 proves an unsigned delegation |
Authentication Enforced and verified
The endpoint asked for a token and published valid RFC 9728 metadata describing how to get one.
| Result | Enforced and verified |
|---|---|
| Enforced | On tool calls |
| HTTP status | 200 |
WWW-Authenticate challenge Bearer resource_metadata="https://mcp.cosend.app/.well-known/oauth-protected-resource/mcp", scope="messages:read messages:write contacts:read contacts:write templates:read templates:write automations:read automations:write runs:read runs:write connectors:read connectors:write usage:read members:read billing:read settings:read connections:read connections:write approvals:read approvals:write campaigns:read campaigns:write"
Bearer resource_metadata="https://mcp.cosend.app/.well-known/oauth-protected-resource/mcp", scope="messages:read messages:write contacts:read contacts:write templates:read templates:write automations:read automations:write runs:read runs:write connectors:read connectors:write usage:read members:read billing:read settings:read connections:read connections:write approvals:read approvals:write campaigns:read campaigns:write" | Header | Value |
|---|---|
| strict-transport-security | max-age=31536000; includeSubDomains |
| content-security-policy | default-src 'none'; frame-ancestors 'none'; base-uri 'none'; form-action 'none' |
| x-content-type-options | nosniff |
| x-frame-options | DENY |
| referrer-policy | no-referrer |
Protected resource metadata
| Document | https://mcp.cosend.app/.well-known/oauth-protected-resource/mcp |
|---|---|
| Retrieved | Yes |
| Resource | https://mcp.cosend.app/mcp |
| Authorisation server | https://auth.cosend.app |
Background: How OAuth 2.1 works in the 2026 MCP spec →
Transports 2 probes
| Transport | URL | Outcome | Status | Location |
|---|---|---|---|---|
| streamable-http | https://mcp.cosend.app/mcp | Verified | 200 | |
| http (plaintext) | http://mcp.cosend.app/mcp | HTTPS enforced | 301 | https://mcp.cosend.app/mcp |
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
acknowledge_category_change Acknowledge a template category change ~68
Marks one template category change, named by id, as seen for the whole workspace, so it leaves the list of open changes. It changes nothing about the template, and nothing is sent to Meta.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The category change to acknowledge, from the list of category changes. |
No output schema declared.
No examples provided.
ask_for_conversation Ask a teammate for a chat ~238
Asks the person who holds one conversation, named by id, in the Cosend Inbox to hand it over; reason is an optional note for them. They get a hand-over prompt: when they say yes, or do not answer within 60 seconds, the chat passes to the person this app acts for, and messages can be sent in it; when they say no, it stays theirs. This is the way into a chat someone else holds: every hand-over starts with asking its holder. A chat nobody holds, or one the person this app acts for already holds, is refused, as there is nothing to ask for and a message can be sent there as it is. Asking again while a request waits returns that request. Returns the request: its state (pending until answered), who holds the chat (held_by) and when it passes without an answer (expires_at). Whether it has passed shows in the conversation’s lock (held_by, can_i_send). The request stays inside Cosend, out of the customer’s view.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The conversation to ask for. |
| reason | string | – | – |
No output schema declared.
No examples provided.
bulk_update_conversations Update many conversations ~238
Applies one action to up to 100 conversations named in ids: resolve marks each Done, assign sets assignee_id (null takes it off), label adds labels_add and takes off labels_remove, priority sets priority, and read marks each read in Cosend. Each chat gets its own result, so one that is not found or that someone else is replying in is skipped with its reason while the rest go through. Each action needs a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve; called again with that token, it acts. Every change stays inside Cosend, out of customers’ view.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| assignee_id | – | – | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| ids | array | yes | – |
| labels_add | array | – | – |
| labels_remove | array | – | – |
| priority | string | – | – |
| read | boolean | – | – |
No output schema declared.
No examples provided.
buy_cosend_number Buy a phone number from Cosend ~339
Buys one local US or Canadian phone number from Cosend for WhatsApp, named by its phone_number and country as the search of available numbers returned them (with its locality and region), and returns it as a Cosend number (cnm_…), usually verifying. It charges the workspace’s card, so it acts only on a confirmed call: without confirmation_token it buys nothing and returns a preview and a token for a person to approve. The preview shows the price of a number, its interval (month or year, as the plan is billed), when it renews and the amount charged today for the rest of the billing period; the price, interval and renewal are read again when the token comes back, and if any changed the token is refused. Changing a single character of the arguments invalidates the token too. Only a person’s approval buys it: the credential alone buys nothing, through this tool or straight to the API. On a paid plan with a card on file, for owners and admins, up to the plan’s number limit; five a minute and ten a day per workspace. The number is on the Numbers page of the Cosend dashboard too.
| Name | Type | Req | Description |
|---|---|---|---|
| category | string | yes | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| country | string | yes | – |
| display_name | string | yes | – |
| locality | string | – | – |
| phone_number | string | yes | – |
| region | string | – | – |
| website | string | – | – |
No output schema declared.
No examples provided.
cancel_campaign Stop a campaign ~90
Stops a campaign, named by id, that is scheduled, sending or paused, and returns it stopped: nothing more goes out, and sends still waiting are cancelled. Messages already sent stay sent. A stopped campaign does not start again; its results stay readable. Stopping a draft, or a campaign that has finished, is refused in words.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
cancel_run Cancel a run ~54
Cancels a run that has not finished, as a stop: it needs no confirmation. Messages the run already sent stay sent. Returns the run as it is after the cancel.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The run id. |
No output schema declared.
No examples provided.
cancel_setup_link Cancel a set-up link ~83
Cancels an open set-up link so its page stops working, for example when the person changed their mind or the link asks for the wrong thing. A link already completed, expired or cancelled is refused, and nothing a link set up is undone. Only the person who asked for it can cancel it.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
change_knowledge Change what the AI assistant knows ~386
Changes one AI assistant’s knowledge, named by id. action add takes a kind: file (media_id of a file uploaded on the assistant’s number), text (text, with an optional title), url (a website) or place (a Google Maps link); it answers at once with the source, and the assistant reads it in the background (a website is read again every month). action remove (sourceId) forgets a source and everything read from it, on a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve. action refresh (sourceId) reads a source again now. action review_suggestion (suggestionId, accepted true or false, optional question and answer edits) turns a suggested answer into knowledge or drops it. Owners and admins.
| Name | Type | Req | Description |
|---|---|---|---|
| accepted | boolean | – | – |
| action | string | yes | add gives the assistant something to read. remove forgets one source. refresh reads one source again. review_suggestion accepts (with edits) or rejects a suggested answer. |
| answer | string | – | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The id in the resource path. |
| kind | string | – | – |
| media_id | string | – | A file uploaded on this automation’s number with POST /v1/media. A file uploaded on another number is refused with media_other_number. |
| question | string | – | – |
| sourceId | string | – | The sourceId in the resource path. |
| suggestionId | string | – | The suggestionId in the resource path. |
| text | string | – | – |
| title | string | – | – |
| url | string | – | – |
No output schema declared.
No examples provided.
change_template Edit, copy or delete a template ~640
Changes an existing message template at Meta. action edit replaces all of its components and returns it to review (an approved template can be edited once in 24 hours); copy creates it on another WhatsApp account of the same business, where Meta reviews it again; delete removes it at Meta and from Cosend once Meta confirms. Each change needs a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve; called again with that token, it acts. Ten changes a day per workspace. It sends no message to anyone. Meta’s reason for a refusal is in Meta’s words.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | edit replaces the template’s components and sends it back to Meta’s review; copy makes a copy of it on another WhatsApp account of the business; delete removes it at Meta. |
| authentication | – | – | An authentication template's choices, as create takes them. An edit is one language version: languages is left out, or exactly the template's own language. |
| category | string | – | A new category, sent to Meta when it differs. Meta refuses a change on an approved template, so that is refused here first. |
| components | array | – | Meta's components, all of them; required on every category but authentication, which takes authentication instead. |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| display_name | string | – | A new human name for the copy, from which its system name is made (with _2, _3… added when the account already has that name). Without it the copy keeps the template’s names, and an account that alre… |
| example_values | object | – | A sample value for every variable in the TEXT header and the body, keyed per component by the name between the braces: {"body": {"1": "Pablo", "2": "860198-230332"}} or {"body": {"first_name": "Pablo… |
| id | string | yes | The template: its tpl_ id, or name:language. |
| language | string | – | The template's language, which an edit keeps. Send the template's own language, or leave it out. |
| name | string | – | The template's system name, which an edit keeps. Send the template's own name, or leave it out. |
| parameter_format | string | – | – |
| waba_id | string | – | The WhatsApp Business Account to copy to, as Cosend's wab_ id (a connection's waba_id, not its meta_waba_id). It must be another account of the same business portfolio. |
No output schema declared.
No examples provided.
chat_notes Add, edit or delete an internal note ~240
Writes the internal notes on one conversation, named by id, visible to the business’s team. action add writes a note (body, with mentions as a list of usr_ ids of members to notify). action edit changes a note’s body (noteId and body; only its author can). action delete removes a note (noteId; its author, an owner or an admin can). Each delete needs a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve; called again with that token, it deletes. add and edit return the note, whose body can quote customers and comes back marked untrusted.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | add, edit or delete. |
| body | string | – | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The id in the resource path. |
| mentions | array | – | – |
| noteId | string | – | The noteId in the resource path. |
No output schema declared.
No examples provided.
check_setup_link Check a set-up link ~184
Reads a set-up link’s status. status open means the step is not done; completed means it has, and result says what was set up; expired means its 30 minutes passed; cancelled means it was cancelled or replaced by a newer one. For connect_integrations, result.connector_accounts lists each account saved; when the workspace has several WhatsApp numbers an account is on none of them until it is added to a number (the Integrations page in Cosend, or the integration account tool’s add_number). For reconnect_number, result names the connection and its number once Meta’s flow finished; for pick_google_file, completed means Google’s picker was opened for the account, and the account’s spreadsheet list shows what was chosen. Only the person who asked for the link can read it. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
contact_fields Add, change or delete a contact field ~260
Writes the workspace’s contact fields: the keys a contact’s attributes can hold. action create adds one (key, label, type text, number, date, boolean or choice, choices for a choice field, required). action update (id) changes a field’s label, type, choices or required; its key stays. action delete (id) removes the field and every contact’s value for it. Creating, changing and deleting are for owners and admins. Each delete needs a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve; called again with that token, it deletes.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | create, update or delete. |
| choices | array | – | A choice field's options, 1 to 100 of them, each once; none otherwise. |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | – | The id in the resource path. |
| key | string | – | – |
| label | string | – | – |
| required | boolean | – | – |
| type | string | – | – |
No output schema declared.
No examples provided.
copy_automation Copy an automation ~112
Makes a custom copy of an existing automation, a recipe’s included, on the same WhatsApp number and switched off, with its own copy of the persona, knowledge and remembered values. name is the copy’s; left out, it is the original’s followed by “(copy)”. The original is unchanged. Returns the copy.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The automation to copy. |
| name | string | – | The copy’s name. Absent: the original’s, followed by “(copy)”. |
No output schema declared.
No examples provided.
create_audience Save an audience for campaigns ~119
Saves an audience for campaigns, with a name, the WhatsApp number it is for and who it names (contact tags and that number’s chat labels, any or all), and returns it with its id, the audience.segment_id a campaign takes. It sends nothing and messages nobody. The audience is on the Campaigns page of the Cosend dashboard too. Campaigns are on the Starter plan and above, and for owners and admins.
| Name | Type | Req | Description |
|---|---|---|---|
| connection_id | string | yes | – |
| definition | object | yes | – |
| name | string | yes | – |
No output schema declared.
No examples provided.
create_campaign Start a campaign draft ~95
Starts a campaign as a draft, with a name and the WhatsApp number it goes out on, and returns it with its id. It sends nothing and messages nobody: a draft waits until a person approves it. The draft is on the Campaigns page of the Cosend dashboard too. Campaigns are on the Starter plan and above, and for owners and admins.
| Name | Type | Req | Description |
|---|---|---|---|
| connection_id | – | – | – |
| name | string | – | – |
No output schema declared.
No examples provided.
create_custom_automation Create a custom automation ~515
Creates a new automation on one WhatsApp number from an automation document (ir) written in full, as Cosend’s automation editor saves one: what it starts on, its steps and its limits, in the document schema that ir_version names. What a document can use on a number, each trigger, step and integration action with what it takes, and that schema itself, are in the read-only list of automation building blocks for the number. The document is checked as it is saved; one that does not pass is refused with each failing path as a JSON pointer. A “Run a script” step runs TypeScript written in the step, in a sandbox with no network and nothing imported; an outside address is reached only by a step that calls an allowlisted URL. A document marked as made from a recipe (meta.created_by "recipe") is refused: a recipe is set up and changed with its own tools. slug, name and connection_id name it and its number. It is created switched off: saving it runs nothing and sends nothing, and it acts only once it is deployed and switched on, each of which names every action its steps take. Its runs count against the plan’s allowance as every automation’s do. Each call needs confirming: without confirmation_token it changes nothing and returns a preview naming every action the document’s steps take, and a token for a person to approve; called again with that token, it saves. Returns the automation, switched off.
| Name | Type | Req | Description |
|---|---|---|---|
| change_note | string | – | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| connection_id | string | yes | The WhatsApp number (connection id) this automation belongs to: its triggers come from it, its messages go from it, and its templates are read on it. Required, and it cannot be moved afterwards. The… |
| description | – | – | – |
| ir | object | yes | The automation’s flow document (the IR), as a JSON object. `ir_version` selects the document schema Cosend validates it against. A document that does not match is refused with `automation_invalid`, w… |
| name | string | yes | – |
| slug | string | yes | – |
No output schema declared.
No examples provided.
create_setup_link Make a set-up link ~425
Makes a link for the person in this chat to open in their own browser, signed in to Cosend, to do one set-up step only they can do there, and returns its url. It works for 30 minutes and only for them, and carries no password or key. kind connect_number connects a WhatsApp number; reconnect_number needs connection_id and runs Meta’s flow again for that number; connect_integrations needs connectors (each integration by its Cosend key, as list_connectors gives them) and completes when every one has an account, and with account_id (and that account’s one connector) the person types a new key for an account that already exists; pick_google_file needs account_id, a Google account, and opens Google’s file picker for it so the person can choose spreadsheets Cosend may open; upgrade needs plan, the plan to buy; payment_method adds a card. The person does the step themselves on the page: they connect the number in Meta’s window, sign in to the integration or type its key, choose files in Google’s picker, pay in Stripe’s Checkout, or add the card. It cannot be done for them. A new link of the same kind cancels the workspace’s open one, whoever made it.
| Name | Type | Req | Description |
|---|---|---|---|
| account_id | string | – | The connected Google account whose file picker opens, by its connector account id. The person chooses the spreadsheets Cosend may open; the link completes when the picker was opened. |
| connection_id | string | – | The WhatsApp number to reconnect, by its connection id. The person runs Meta’s flow again for it, and the link completes when that number is connected. |
| connectors | array | – | The integrations the person signs in to, by key from GET /v1/connectors. The link completes when every one of them has an account. |
| kind | string | yes | – |
| plan | string | – | The plan to buy, by code from GET /v1/billing/plans. Only for a workspace with no paid subscription, and only to a higher plan. |
No output schema declared.
No examples provided.
create_spreadsheet Make a spreadsheet ~191
Makes a spreadsheet in a connected Google account, named by id: its title, one tab, the header columns (with a type each, text by default) and, optionally, one example row. It lands in the account’s own Google Drive and answers with its id, link, tabs and columns, which an automation’s sheet steps then use. It only creates a new spreadsheet in Google Drive.
| Name | Type | Req | Description |
|---|---|---|---|
| columns | array | yes | The header row, first column first: 26 at most, A to Z. |
| example_row | array | – | One row under the header for the owner to type over, as wide as `columns`. Absent: the sheet has the header row only. |
| id | string | yes | The id in the resource path. |
| tab | string | yes | The one tab’s name. |
| title | string | yes | The spreadsheet’s name, in the owner’s words. |
No output schema declared.
No examples provided.
delete_automation Delete an automation ~131
Deletes an automation. One that is switched on is switched off with it, no run starts after it, it leaves the list of automations, and its id stops answering. Each call needs confirming: without confirmation_token it changes nothing and returns a preview and a token for a person to approve; called again with that token, it deletes.
| Name | Type | Req | Description |
|---|---|---|---|
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The automation to delete. |
No output schema declared.
No examples provided.
delete_contact Delete a contact ~242
Deletes one contact, named by id. Without purge the contact and its chats drop out of lists, counts and the Inbox, and Cosend keeps their messages until a purge; if the person writes again, they become a new contact with a new chat. With purge true the contact and what Cosend holds about them are erased for good (an owner’s action, for a data-protection request). The deletion stays inside Cosend, out of the contact’s view. Each delete needs a confirmed call: without confirmation_token the call is a preview that names the contact and returns a token for a person to approve; called again with that token, it deletes.
| Name | Type | Req | Description |
|---|---|---|---|
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The id in the resource path. |
| purge | boolean | – | true to erase the contact for good, as a data-protection erasure does (owners only); omit to take the contact and its chats out of lists and the Inbox, with its messages kept. |
No output schema declared.
No examples provided.
deploy_automation Deploy an automation ~350
Compiles one version of an automation, runs Cosend’s safety checks on it and stores the result, so it can be switched on. It does not switch the automation on, and nothing is sent. version_id is the version; left out, the newest. actions names every action the version’s steps take, and those of a second automation set up with it, from the list in the schema (an automation’s versions list theirs as actions); a list that leaves one out or names another is refused, with the right list. The preview refuses that list as the call would, and gives no token. confirm_actions names, one by one, each destructive or money-moving action the person approved; one left unnamed is refused by name. Each call needs confirming: without confirmation_token it changes nothing and returns a preview and a token for a person to approve; called again with that token, it deploys. A confirmed deploy that fails to compile uses up the confirmation.
| Name | Type | Req | Description |
|---|---|---|---|
| actions | array | yes | Every action the automation’s steps take, and those of a second automation set up with it, each once: what a person approves it doing. A list that leaves one out, or names one the steps do not take,… |
| confirm_actions | array | – | – |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The id in the resource path. |
| version_id | string | – | – |
No output schema declared.
No examples provided.
get_approvals Read approvals ~225
Reads the human-decision queue. view list returns approvals newest first, one page at a time, optionally filtered by state; a member sees the queue of their own numbers. view one returns one approval with exactly what the human is shown (preview) and the arguments of the held action. view action_types returns every action type that can require approval with its default risk. Previews and arguments can quote customers and come back marked untrusted. A person sets the workspace’s mode per type, and decides each approval, in the Cosend dashboard. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| id | string | – | The id in the resource path. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| state | string | – | With view list: only approvals in this state. |
| view | string | yes | list lists approvals. one reads one approval by id. action_types lists what can need approval. |
No output schema declared.
No examples provided.
get_assistant_persona Read the AI assistant’s persona ~108
Reads the persona of one AI assistant, named by id (an automation of this workspace that has an AI step): business name, what the business does, tone, whether it may quote prices, the links it may send and when, the words it avoids, and the confidence threshold below which it hands a chat to a person. An assistant nobody has edited reads as the defaults, with updated_at null. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
get_automation Read one automation ~222
Reads one automation in one of three views. details: its name, status, number, the live version, what it clashes with while it is on and, on request, its versions and recipe settings. versions: every version, newest first, one page at a time. stats: the chats it answered alone and handed over, by day, with the top hand-over reasons. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| expand | string | – | Comma-separated extras for view details: active_version, latest_version, ir, settings. |
| id | string | yes | The automation id. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| range | string | – | For view stats: today or 7d. Defaults to today. |
| view | string | yes | details for the automation itself; versions for every version, newest first; stats for how it did today or over seven days. |
No output schema declared.
No examples provided.
get_billing Read plan and invoices ~153
Reads the workspace’s billing in one of four views. subscription: plan, status, billing period and whether payment is overdue. plans: the plans it may move to, in order, with prices, allowances, seats and features. invoices: when, how much and whether paid, with Stripe’s own document. invoice: one of them. Paying and changing a card happen on a set-up link. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| invoiceId | string | – | The invoice’s Stripe id, from the invoice list (view invoice). |
| view | string | yes | subscription for the current plan, status and period; plans for the plans this workspace may move to; invoices for every invoice; invoice for one, named by invoiceId. |
No output schema declared.
No examples provided.
get_business_profile Read a number’s business profile ~86
Reads the WhatsApp business profile customers see for one number, named by id, from Meta: about, description, address, email, websites, industry and the picture’s address. read_only_reason says when the profile is edited elsewhere (a coexistence number’s is edited in the WhatsApp Business app). Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
get_campaign Get a campaign and its results ~128
Returns one campaign, named by id, and its results together. The campaign: its state, number, template and the value of each blank, saved audience, time, quiet hours, and why it paused, stopped or could not start. The results: how many it was for, sent, delivered, read, replied, failed, were left out and are still to go, failures by group, sends per 10 minutes and what is left by day. A reply counts for 72 hours after the campaign ends. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
get_contact Open a contact ~220
Reads one contact, named by id (a ct_ id, an E.164 phone number or a BSUID), in one of four views. view contact returns the contact: names, phone number or BSUID, tags, attributes, time zone and whether it is blocked. view consent returns the consent state for marketing, utility and authentication, with the entry behind each (where it came from, when, who recorded it, the sentence the person agreed to) and Meta’s own marketing opt-out. view window returns whether a free-form message can be sent now and when the 24-hour customer service window closes. view chats returns this person’s conversation on each number you can see, newest first. Names, attributes and consent sentences come from outside this business and come back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
| view | string | yes | contact for the contact itself; consent for consent per category with its evidence; window for the messaging window; chats for this person’s chats on each number. |
No output schema declared.
No examples provided.
get_conversation Open a conversation ~256
Reads one conversation, named by id. view conversation returns the chat itself: its contact and newest message, its state, who it is assigned to, its labels, priority and snooze, the hand-over summary when a person is needed, who holds the reply lock, and the 24-hour customer service window (open or closed, and when it closes). view activity returns the chat’s activity feed one page at a time: hand-overs, assignments, labels, Done and reopened, optionally only the kinds named in kind. Names, message text and hand-over summaries were written by people outside this business, or from what they wrote, and come back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| id | string | yes | The id in the resource path. |
| kind | array | – | Activity view only: keep only these kinds of activity. Repeat for several. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| view | string | yes | conversation for the chat itself, with its window and who holds it; activity for its activity feed. |
No output schema declared.
No examples provided.
get_data_policy Read how long data is kept ~36
Returns what the workspace keeps, in which group and for how long, and any queued erasure of the whole workspace. Read-only.
Input schema present but exposes no named parameters.
No output schema declared.
No examples provided.
get_inbox_overview See the Inbox at a glance ~251
Reads the figures the Inbox and Home show, one part at a time. part counts returns how many chats are waiting for a person, assigned to the caller, unread, in all (Done and snoozed ones included), and have a 24-hour window closing within four hours, per number and in total, with the approvals waiting. part labels returns each label in use on each number with how many open chats carry it. part today returns today’s replies by who sent them (the assistant, people, the phone, automations) and the chats handed to a person, marked Done and opened, and the AI receptionist’s chats (handled, answered alone, handed over, as its page and Reports count them), in the workspace’s time zone. part week returns replies and new chats for each of the last 7 days. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| connection_id | array | – | Only this WhatsApp number; repeat it for several. Omit for every number you can see. |
| part | string | yes | counts for chats waiting, mine, unread and windows closing; labels for the labels in use; today for today’s replies by who sent them; week for the last 7 days. |
No output schema declared.
No examples provided.
get_integration_accounts See connected integration accounts ~212
Lists the workspace’s connected integration accounts, one page at a time, or reads one when id is given. Each has its integration, name, status, the WhatsApp numbers it is used on and the reason it last failed, if it did. Credentials stay sealed in Cosend. Labels and error text come from the provider and are marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| connection_id | string | – | Only the accounts used on this WhatsApp number. |
| connector | string | – | Only accounts of this integration, by its Cosend key. |
| id | string | – | One integration account to read. Left out, the accounts are listed. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| status | string | – | Only accounts in this state: active, needs_reauth, error, disabled or revoked. |
No output schema declared.
No examples provided.
get_knowledge Read what the AI assistant knows ~234
Reads one AI assistant’s knowledge, named by id. view sources lists every source with its kind (file, text, website, Google place), status (queued, ready, failed), the reason a read failed, the website pages it could not load, and when it was last read; it pages with after and limit. view summary returns what the assistant learned: a short summary, the business name and one-line description read from the material, and suggested questions and answers awaiting a decision. Titles, summaries, questions and answers were read from websites and files by Cosend and come back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| id | string | yes | The id in the resource path. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| view | string | yes | sources lists the files, pasted texts, websites and Google places the assistant reads. summary reads “what I learned”: a short summary and the suggested questions and answers. |
No output schema declared.
No examples provided.
get_media Open a file from a chat ~83
Reads one media file, named by its med_ id (a message’s media block carries it): its type, file name, size and state, and a signed link to download it that works for 15 minutes. File names come from people outside this business and come back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
get_message Open one message ~92
Reads one message, named by its msg_ id or by the WhatsApp message id (wamid): its type and content, direction, delivery status with the time of each step, who sent it, and any file, location, buttons or reactions it carries. Message text comes from people outside this business and comes back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The id in the resource path. |
No output schema declared.
No examples provided.
get_number Read one WhatsApp number ~168
Reads one WhatsApp number in one of four views. summary: its status, quality, limits and onboarding summary. import_progress: the progress of importing the number’s chat history. connect_steps and attempt_steps: an ordered list of the steps of connecting it, each with its state, time and outcome, which explains where a connect that stopped stopped. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The WhatsApp number (connection) id; for view attempt_steps, the connect attempt id. |
| view | string | yes | summary for the number with its onboarding summary; import_progress for how far the chat-history import has come; connect_steps for the steps of how the number was connected, with their times and out… |
No output schema declared.
No examples provided.
get_reply_order See the order a number’s automations answer in ~59
Returns the order in which one WhatsApp number’s automations answer a customer’s message. The order follows from the automations themselves. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The WhatsApp number (connection) id whose automations are ordered. |
No output schema declared.
No examples provided.
get_report Read a Reports card ~324
Reads one card of Reports for a period: how fast chats were answered by the AI receptionist and by people (response), what the receptionist answered alone and handed over with the top reasons (receptionist), each number’s chats (numbers), each member’s chats, response times and chats marked Done (team), chats by hour of the day (hours), or each Click-to-WhatsApp ad’s chats, replied to and handed to a person (ads). Owners and admins, on the Starter plan and above. Each card covers the period asked for, from the first day Reports recorded. Hand-over reasons, member names and ad words come back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| card | string | yes | response for how fast chats were answered; receptionist for what the AI receptionist handled; numbers for each WhatsApp number; team for each member; hours for chats by hour of the day; ads for each… |
| connection_id | array | – | Only this WhatsApp number; repeat it for several. |
| from | string | – | A custom period’s first day (YYYY-MM-DD), with to. |
| kind | string | – | For card response: first (the default) for each chat’s first answer, every for every answer. |
| period | string | – | today, 7d (the default), 30d, or a calendar month as month:YYYY-MM, in the workspace’s time zone. Not with from and to. |
| to | string | – | A custom period’s last day (YYYY-MM-DD), with from. |
No output schema declared.
No examples provided.
get_run Read an automation run ~139
Reads one automation run and its step-by-step trace: under run, the run’s status, timings, what it sent and why it failed; under steps, each step it took with its status and error, one page at a time. Text inside a run can quote what a customer wrote, and comes back marked untrusted. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| id | string | yes | The id in the resource path. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
No output schema declared.
No examples provided.
get_setup_guide Read the set-up guide ~52
Returns the workspace’s set-up guide: each step with its state, done, to do or skipped, worked out from what the workspace already has. It shows what is left to set up. Read-only.
Input schema present but exposes no named parameters.
No output schema declared.
No examples provided.
get_template Read one template ~82
Reads one message template: its components, category, language, Meta’s review status, whether it can be sent right now and, for a rejected one, Meta’s reason in Meta’s own words (marked untrusted). Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | The template: its tpl_ id, or name:language (for example order_update:en_US). |
No output schema declared.
No examples provided.
get_usage See plan usage ~43
Reports this workspace’s usage for one billing period against what its plan includes. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| period | string | – | Which billing period. Omit for the current one. |
No output schema declared.
No examples provided.
import_contacts Import contacts from a CSV ~534
Imports contacts from a CSV file for one WhatsApp number, in steps. step start uploads the file (connection_id; data_base64, or media_id of a text or CSV file already uploaded; file_name) and returns its id, columns, first rows and row count. step map (id) sets mapping (each column as phone, name, tags, locale, timezone, skip or field:key; name and tags may be on several columns), new_fields (fields the run adds to the workspace, each key, label, type and choices, for columns mapped to field:key), connection_id, and consent: the opt-in categories everyone in the file gave, the sentence they agreed to and who attests it. step check (id) reports what a run would do over the whole file: new, updated, invalid, duplicates, already opted out and consent recorded, and with details true the rows left out and five contacts as they will be (a phone typed without its country code takes the country of the number); it writes nothing. step run (id) imports in the background and returns at once; the import’s progress shows in the list of imports. Owners and admins run imports. Each run needs a confirmed call: without confirmation_token the call is a preview that shows the import, its mapping, its new fields and its consent, and returns a token for a person to approve; called again with that token, it runs. The import stays inside Cosend, out of the contacts’ view. File contents come from outside this business and come back marked untrusted.
| Name | Type | Req | Description |
|---|---|---|---|
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| connection_id | string | – | – |
| consent | – | – | – |
| data_base64 | string | – | The CSV file, base64-encoded. |
| details | boolean | – | Also answer the rows that will not be imported (the first 100) and five contacts as the import will write them. |
| file_name | string | – | – |
| id | string | – | The id in the resource path. |
| mapping | object | – | – |
| media_id | string | – | A med_ id from POST /v1/media (text/plain or text/csv), instead of data_base64. |
| new_fields | array | – | – |
| step | string | yes | start to upload the file; map to set the columns, the number and the consent; check to see what a run would do; run to import. |
No output schema declared.
No examples provided.
inbox_views Save, change or delete an Inbox view ~220
Writes the saved views in the Inbox’s Views menu: a name and the conversation filters it applies. action save makes a view (name, filters, and shared: private or workspace). action update (id) renames a view or changes its filters or who sees it; a shared view can be changed by its creator, an owner or an admin. action delete (id) removes a view, on the same terms. Each delete needs a confirmed call: without confirmation_token the call is a preview that returns what would happen and a token for a person to approve; called again with that token, it deletes.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | save, update or delete. |
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| filters | object | – | – |
| id | string | – | The id in the resource path. |
| name | string | – | – |
| shared | string | – | – |
No output schema declared.
No examples provided.
keep_cosend_number Keep a stopped phone number from Cosend ~220
Keeps a phone number from Cosend that was stopped, named by id (cnm_…), before its last day (release_at), and returns it ready: it is not given back, and from the plan’s next renewal it is billed with the plan again, at $5.95 a month, or $59.50 a year when the plan is billed yearly. Nothing is charged today. It acts only on a confirmed call: without confirmation_token it changes nothing and returns a preview of the number and a token for a person to approve; the number’s status, last day and WhatsApp connection are read again when the token comes back. Only a person’s approval keeps it. On a paid plan, for owners and admins.
| Name | Type | Req | Description |
|---|---|---|---|
| confirmation_token | string | – | For a call that needs approval: left out, the call does nothing and returns a preview and a token; given that token, with the same arguments, the call does it. |
| id | string | yes | The stopped Cosend number (cnm_…) to keep. |
No output schema declared.
No examples provided.
list_audiences List saved audiences for campaigns ~152
Lists the saved audiences of this workspace, one page at a time: each one’s id, name, the WhatsApp number it was saved for, and who it names (contact tags and that number’s chat labels, any or all). A campaign’s audience is one of these. They are saved on the Campaigns page of the Cosend dashboard, or by an AI app, by an owner or an admin; this tool only reads them. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
No output schema declared.
No examples provided.
list_automation_building_blocks List automation building blocks ~156
Lists what a custom automation on one WhatsApp number can be built from: each thing it can start on, with the settings it takes and the conditions it can start on; each step it can take, with the inputs it takes and what it returns; each integration working for that number, with its actions and their inputs; and the JSON Schema of the automation document, with the operators and functions a condition is written in. Only what the workspace’s plan includes and the number’s working integrations allow is listed. Read-only: it runs nothing and changes nothing.
| Name | Type | Req | Description |
|---|---|---|---|
| connection_id | string | yes | The WhatsApp number (connection id) the automation is for: only what can be built and run on it is listed. |
| Name | Type | Req | Description |
|---|---|---|---|
| blocks | array | yes | – |
| document_schema | object | – | The JSON Schema of an automation document (`ir_version` 1.0): what it starts on, its steps (actions, conditions, branches, loops, waits) and the expressions a condition is written in, each operator a… |
| integrations | array | yes | Every integration that has actions, with them: what a “Call an integration” step can name as its `integration` and `action`, and what each takes and answers. |
| object | string | yes | – |
| triggers | array | yes | – |
| version | string | yes | – |
No examples provided.
list_automation_recipes List automation recipes ~138
Lists Cosend’s automation recipes as cards, one page at a time: each one’s slug, name, what it does, its category, the settings it asks for and the integrations it needs. Each recipe is set up with a tool of its own named after its slug. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| connection_id | string | – | Marks the recipes that cannot work on this WhatsApp number. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
No output schema declared.
No examples provided.
list_automations List automations ~94
Lists this workspace’s automations, one page at a time, with each one’s status, the version that is live, and today’s runs and failures. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
No output schema declared.
No examples provided.
list_campaigns List campaigns ~198
Lists this workspace’s campaigns, newest first, one page at a time: each one’s id, state (draft, scheduled, sending, paused, sent or stopped), number, template, saved audience, time, quiet hours, and how many it was for, sent, delivered, read, got a reply from and failed for. These are the campaigns on the Campaigns page of the Cosend dashboard. Read-only.
| Name | Type | Req | Description |
|---|---|---|---|
| after | string | – | The next_cursor of the previous page, to read the page after it. Omit for the first page. |
| connection_id | array | – | Only campaigns on this WhatsApp number. Repeat for several. |
| limit | integer | – | How many records to return. Defaults to 25, maximum 100. |
| q | string | – | Only campaigns whose name contains this. |
| status | array | – | Only campaigns in this state, e.g. draft or sending. Repeat for several. |
No output schema declared.
No examples provided.
How do I install the Cosend MCP server?
Cosend is a hosted endpoint at https://mcp.cosend.app/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
remote · mcp.cosend.app
claude mcp add --transport http app-cosend-mcp 'https://mcp.cosend.app/mcp'
What is the Cosend MCP server?
Cosend is an MCP server listed in the public MCP registry as app.cosend/mcp. WhatsApp Business inbox for AI apps: read and answer chats, run campaigns, deploy automations. This page covers its hosted endpoint (https://mcp.cosend.app/mcp).
Is the Cosend MCP server safe to use?
Cosend scores 84 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the Cosend MCP server expose?
Cosend exposes 125 tools: whoami, list_conversations, search_conversations, read_conversation, send_message, and 120 more. Their descriptions and schemas cost roughly 29,877 tokens of context every time the server is loaded.
Does the Cosend MCP server require authentication?
Yes. Cosend asked us for credentials when we connected, so you will need to authorise it in your MCP client before it can do anything.
Is the Cosend MCP server still maintained?
Cosend is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What did VerifyMCP test on the Cosend MCP server?
VerifyMCP ran 13 checks on Cosend, covering TLS certificate, Authorization, HTTPS, HSTS header and 9 more. Supply Chain Security and Provenance & Transparency do not apply to a hosted endpoint, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.