Skip to content

Connectors · AI

Deepgram MCP server for AI agents.

actionlake is an MCP gateway: one remote MCP server that connects Claude and other MCP clients to Deepgram and the rest of your systems. It gives your agent 50 Deepgram actions (28 reads, 14 writes and 8 that wait for a confirmation) and holds the Deepgram credential itself, so the agent never does. It paces Deepgram's limits and records every call.

100,000 free calls a month. No card, no email, no password: press Start a free sandbox on the next screen and keep the key it shows you.

claude mcp add --transport http actionlake https://connect.actionlake.com/mcp
the first call
do {  action: "deepgram.projects.list",  intent: "check which Deepgram account is connected"}# List Projects · read · 1 upstream call
  • 50actionsin one signed file
  • 28readlook, never change
  • 14writecreate and update
  • 8wait for a yesdestructive, confirmed first

Set up

Connect Deepgram in three steps.

  1. 01

    Add actionlake to Claude.

    Run the command once in Claude Code; its consent page can create a free sandbox: 100,000 calls a month, no card. In the Claude app, add a custom connector with the URL below and paste your key on the consent page. Creating a new sandbox from the Claude app is not yet tested by us.

    https://connect.actionlake.com/mcp
  2. 02

    Connect Deepgram.

    API key. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead. actionlake stores what you paste write-only, sealed under your tenant's key.

    • api_keyDeepgram API key (Console > Settings > API Keys)
    • hostWhere your Deepgram lives: your host
    account { op: "connections.requirements", service: "deepgram" }
    # free: exactly what to paste, and what is found for you
  3. 03

    Make the first call. Check before a write.

    Ask in plain words; Claude picks the action. Before anything changes, check validates the call and prices it, for free.

    do    { action: "deepgram.projects.list" }# list projects · read · 1 upstream callcheck { action: "deepgram.keys.create", input: { … } }# free: inputs, permissions, cost and confirmations, nothing sent

Deepgram at a glance

Sign-in
API key. No consent screen: your agent can connect it for you, but the key then passes through your chat. To keep it out of the model's context, paste it in the console instead.
You provide
api_key
Setup fields
host
Limits we pace to
100 calls a minute, per connection
API covered
Deepgram REST API 1.0.0api.deepgram.com
Built from
Deepgram's documentation, read 22 October 2026developers.deepgram.comdevelopers.deepgram.com
Service file
Version 1.0.0, signed (Ed25519)
Cached reads
20 of 28 reads, for up to 5 minutes; a write clears what it affects
Shared entities
3, so ask can line Deepgram up with your other systems
Data
Processed and stored in Sydney, Australia. No US region yet

Capabilities

What an AI agent can do in Deepgram.

50 actions in 21 areas. Every one carries its effect, so a key can allow the reads and hold the rest.

  • 28readlook, never change
  • 14writecreate and update
  • 8destructivewait for a confirmation
  1. Agent configurations

    5

    2 read · 2 write · 1 destructive

    Create an Agent Configuration · List Agent Configurations · Get an Agent Configuration · Update Agent Metadata

  2. Agent variables

    5

    2 read · 2 write · 1 destructive

    Create an Agent Variable · List Agent Variables · Get an Agent Variable · Update an Agent Variable

  3. Projects

    5

    2 read · 1 write · 2 destructive

    List Projects · Get a Project · Update a Project · Delete a Project

  4. Distribution credentials

    4

    2 read · 1 write · 1 destructive

    Create a Project Self-Hosted Distribution Credential · List Project Self-Hosted Distribution Credentials · Get a Project Self-Hosted Distribution Credential · Delete a Project Self-Hosted Distribution Credential

  5. Keys

    4

    2 read · 1 write · 1 destructive

    List Project Keys · Create a Project Key · Get a Project Key · Delete a Project Key

  6. Invites

    3

    1 read · 1 write · 1 destructive

    Create a Project Invite · List Project Invites · Delete a Project Invite

  7. Usage

    3

    3 read · 0 write · 0 destructive

    Get Project Usage · Get Project Usage Breakdown · List Project Usage Fields

  8. Balances

    2

    2 read · 0 write · 0 destructive

    Get Project Balances · Get a Project Balance

13 more areas · 19 actions
  • Billing 2
  • Member scopes 2
  • Members 2
  • Models 2
  • Project models 2
  • Requests 2
  • Agent think models 1
  • Listen 1
  • Purchases 1
  • Read 1
  • Speak 1
  • Speak V2 1
  • Tokens 1

Top actions

The Deepgram API, shaped for agents.

The ids your agent types, what each does, and the one name it shares with other systems. 7 of Deepgram's 7 action names are shared with at least one other connector, so an agent learns them once.

ActionWhat it doesEffectSame name in
deepgram.keys.listList Project KeysList Project Keysreadidentity.api_key.list96 connectors
deepgram.keys.createCreate a Project KeyCreate a Project Keywriteidentity.api_key.create100 connectors
deepgram.invites.createCreate a Project InviteCreate a Project Invitewriteidentity.invitation.create41 connectors
deepgram.invites.listList Project InvitesList Project Invitesreadidentity.invitation.list33 connectors
deepgram.keys.deleteDelete a Project KeyDelete a Project Keydestructiveidentity.api_key.delete76 connectors
deepgram.members.listList Project MembersList Project Membersreadthis connector
deepgram.projects.listList ProjectsList Projectsreadthis connector
deepgram.models.listList ModelsList Modelsreadthis connector
deepgram.speak.createText to Speech transformationText to Speech transformationwritethis connector
deepgram.balances.listGet Project BalancesGet Project Balancesreadthis connector
deepgram.requests.listList Project RequestsList Project Requestsreadthis connector
deepgram.speak_v2.createFlux Text to Speech (batch)Flux Text to Speech (batch)writethis connector

Try it

Prompts to paste into Claude.

Each one names the Deepgram actions it will call, so you know what it reads and what it changes before you send it.

  1. Readread

    Give me a one-line status for every active Deepgram project.

    callsdeepgram.projects.list

  2. Writewrite

    Create a Project Key in Deepgram. Show me exactly what you will send before you save it.

    callsdeepgram.keys.create

Governance for Deepgram

Destructive actions wait for a yes.

8 Deepgram actions are destructive. These are all of them:

  • Delete a Project Keydeepgram.keys.delete
  • Delete a Project Invitedeepgram.invites.delete
  • Delete a Project Memberdeepgram.members.delete
  • Delete a Projectdeepgram.projects.delete
  • Delete an Agent Variabledeepgram.agent_variables.delete
  • Delete an Agent Configurationdeepgram.agent_configurations.delete
  • Delete a Project Self-Hosted Distribution Credentialdeepgram.distribution_credentials.delete
  • Leave a Projectdeepgram.projects.leave

A destructive action runs only with confirm: true. Under the human policy the agent cannot confirm for itself: the run stops until a named person approves, in the console or from a signed link in an email. The cap on a key is refused before the call is sent, never after.

dodeepgram.keys.deleteconfirm_policy: human

Delete a Project Key?

intent: the user asked me to delete a Project Key

run suspended · waiting for a named person · key deepgram-agent cannot confirm itself
In the console, or from an emailed signed link.
a key for this connector
account {  op: "keys.create", name: "deepgram-agent",  allowed_actions: ["deepgram.*"], cap_calls: 2000,  confirm_policy: "human_for_writes",  approval_rules: [    { match: "deepgram.keys.delete", approver: "[email protected]" },    { match: "deepgram.invites.delete", approver: "[email protected]" },    { match: "deepgram.members.delete", approver: "[email protected]" }  ]}
Illustrative. Scoped to Deepgram only, capped at 2,000 upstream calls, every write confirmed, and the destructive ones approved by a named person.

Cost

What Deepgram costs on actionlake.

Every Deepgram action is one upstream call, the only thing actionlake bills. Searching, check, cost, account and any call Deepgram refused with 401, 403, 429 or a 5xx are free. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000.

  • 1upstream call per actionall 50 Deepgram actions
  • 14actions pageeach page of a long list is one call
  • 20reads can be cacheda cache hit costs nothing
  • 100,000calls a month, freepaid plans from US$49 a month for 250,000

See pricing

Action index

Every Deepgram action.

50 actions by area: the id your agent calls, what it does, and its effect.

Agent configurations5
  • deepgram.agent_configurations.createCreate an Agent Configurationwrite
  • deepgram.agent_configurations.deleteDelete an Agent Configurationdestructive
  • deepgram.agent_configurations.getGet an Agent Configurationread
  • deepgram.agent_configurations.listList Agent Configurationsread
  • deepgram.agent_configurations.updateUpdate Agent Metadatawrite
Agent variables5
  • deepgram.agent_variables.createCreate an Agent Variablewrite
  • deepgram.agent_variables.deleteDelete an Agent Variabledestructive
  • deepgram.agent_variables.getGet an Agent Variableread
  • deepgram.agent_variables.listList Agent Variablesread
  • deepgram.agent_variables.updateUpdate an Agent Variablewrite
Projects5
  • deepgram.projects.deleteDelete a Projectdestructive
  • deepgram.projects.getGet a Projectread
  • deepgram.projects.leaveLeave a Projectdestructive
  • deepgram.projects.listList Projectsread
  • deepgram.projects.updateUpdate a Projectwrite
Distribution credentials4
  • deepgram.distribution_credentials.createCreate a Project Self-Hosted Distribution Credentialwrite
  • deepgram.distribution_credentials.deleteDelete a Project Self-Hosted Distribution Credentialdestructive
  • deepgram.distribution_credentials.getGet a Project Self-Hosted Distribution Credentialread
  • deepgram.distribution_credentials.listList Project Self-Hosted Distribution Credentialsread
Keys4
  • deepgram.keys.createCreate a Project Keywrite
  • deepgram.keys.deleteDelete a Project Keydestructive
  • deepgram.keys.getGet a Project Keyread
  • deepgram.keys.listList Project Keysread
Invites3
  • deepgram.invites.createCreate a Project Invitewrite
  • deepgram.invites.deleteDelete a Project Invitedestructive
  • deepgram.invites.listList Project Invitesread
Usage3
  • deepgram.usage.getGet Project Usageread
  • deepgram.usage.get_breakdownGet Project Usage Breakdownread
  • deepgram.usage.get_fieldsList Project Usage Fieldsread
Balances2
  • deepgram.balances.getGet a Project Balanceread
  • deepgram.balances.listGet Project Balancesread
Billing2
  • deepgram.billing.get_breakdownGet Project Billing Breakdownread
  • deepgram.billing.get_fieldsList Project Billing Fieldsread
Member scopes2
  • deepgram.member_scopes.listList Project Member Scopesread
  • deepgram.member_scopes.updateUpdate Project Member Scopeswrite
Members2
  • deepgram.members.deleteDelete a Project Memberdestructive
  • deepgram.members.listList Project Membersread
Models2
  • deepgram.models.getGet a specific Modelread
  • deepgram.models.listList Modelsread
Project models2
  • deepgram.project_models.getGet a Project Modelread
  • deepgram.project_models.listList Project Modelsread
Requests2
  • deepgram.requests.getGet a Project Requestread
  • deepgram.requests.listList Project Requestsread
Agent think models1
  • deepgram.agent_think_models.listList Agent Think Modelsread
Listen1
  • deepgram.listen.transcribe_urlTranscribe and analyze pre-recorded audio and videowrite
Purchases1
  • deepgram.purchases.listList Project Purchasesread
Read1
  • deepgram.read.analyzeAnalyze text contentwrite
Speak1
  • deepgram.speak.createText to Speech transformationwrite
Speak V21
  • deepgram.speak_v2.createFlux Text to Speech (batch)write
Tokens1
  • deepgram.tokens.grantToken-based Authenticationwrite

Questions

The Deepgram MCP server, answered.

Is there a Deepgram MCP server for Claude?

Yes. actionlake is an MCP gateway: a remote MCP server that gives Claude, or another MCP client, 50 Deepgram actions across agent configurations, agent variables, projects and distribution credentials, through the same eight tools it uses for every other connector. actionlake is independent; it is not made or endorsed by Deepgram.

What do I need to connect Deepgram?

Deepgram API key (Console > Settings > API Keys). Plus your host. The free call account { op: "connections.requirements", service: "deepgram" } lists exactly what to paste.

Can an AI agent delete or change things in Deepgram?

Only if its key allows it, and only with an explicit confirmation. 8 Deepgram actions are destructive, including delete a Project Key, delete a Project Invite and delete a Project Member. Each runs only with confirm: true. Under the human confirm policy, or an approval rule on the key, the run waits until a named person approves it.

Will an AI agent use up my Deepgram API limit?

actionlake paces every call to the limits in the Deepgram service file: 100 calls a minute per connection. Repeated reads are answered from cache (20 Deepgram actions can be), which never reaches Deepgram.

What does a Deepgram action cost?

One upstream call, the only unit actionlake bills. A list that pages bills one call per page (14 Deepgram actions page). Cache hits, searches, dry runs and calls Deepgram refused with 401, 403, 429 or a 5xx are free. The Free plan includes 100,000 calls a month with no card. Paid plans start at US$49 a month for 250,000 calls, with overage from US$0.30 down to US$0.18 per 1,000.

Which Deepgram API does it cover?

The Deepgram REST API 1.0.0 at api.deepgram.com by default, or the base URL set when you connect. It is built from Deepgram's documentation as read on 22 October 2026. 50 actions in 21 areas, each one in the signed service file (version 1.0.0).

Does actionlake keep my Deepgram data, and where?

actionlake processes and stores data in Sydney, Australia; there is no US region yet. Reads may be cached for up to 5 minutes, and a write clears the reads it affects. Each run's request and response is kept, sealed, for your plan's trace window: 1 day on Free, 3 days on Starter, 7 days on Pro, 14 days on Team, 30 days on Business and 30 days on Enterprise. Your Deepgram credential is sealed under your tenant's key, never logged and never returned. Your traffic is not used to train models.

Which AI clients can use it?

Claude Code, tested by us end to end, and the Claude app as a custom connector, tested by us with an existing key. Other MCP clients that support remote servers over Streamable HTTP (MCP revisions 2025-06-18 and 2025-11-25) and OAuth dynamic client registration, or a bearer header, should connect the same way; we have not tested them. Or plain REST, under the same action names.

Give your agent Deepgram today.

Start free - 100,000 calls/month, no card. No signup. Hard caps on every key. Cached answers are free.