Claude Render connectorWhat Claude can do in your Render account.
The Claude Render connector exposes 25 tools. 18 read your hosting (services, logs, metrics, databases) and 7 create or change something in it. Here is what Claude can check, what it can launch, and what it sends back to the dashboard.
Verified Trustpilot reviews · AI, automation & growth agency
What the Claude Render connector changes for your team
When a service goes down, you no longer need to open the Render dashboard to find out why. You ask in the chat, Claude reads the logs, metrics or deploy history, and answers in plain English. It can also create a resource or kick off a deploy, inside a scope Render has deliberately kept narrow.
Diagnose an outage without digging. A customer reports errors: list_logs surfaces the filtered log lines, get_metrics shows the load at that moment, and list_deploys tells you whether a recent release lines up.
Stand up infrastructure in one sentence. For a prototype, create_web_service, create_postgres or create_key_value set up the service, database or cache without the forms.
Ask a database a question. query_render_postgres runs a read-only SQL query on a Render-hosted database, and Claude explains what came back.
What Claude cannot do here: it deletes nothing, it does not touch scaling, and on an existing service it only changes two things, the environment variables and the start of a deploy. The three tools named update_ for a web service, static site or cron job only return a dashboard link. Nothing runs on its own either: no tool reacts to an incident unless you ask. If you want background monitoring, that is a job for an automation platform, a different kind of tool; the Integrations hub covers that side.
The vocabulary in one minute
Five words worth knowing before you connect Render.
- Connector
- The link you set up once between Claude and your Render account, so Claude can work in it while answering you.
- Tool
- One named action the connector gives Claude. Claude decides on its own which to call; the directory sheet lists them by name.
- Authorization
- Render's own sign-in screen, where you grant Claude the access it uses afterwards. Granted once per person, and revocable the same way.
- Approval
- The confirmation Claude waits for before carrying out something that changes your account, shown in the chat when it matters.
- MCP
- The open standard connectors are built on; it is what lets an assistant like Claude talk to an outside service.
Connect Render to Claude in three steps
- 01
Find Render in settings
Open Claude's settings, go to Customize then Connectors, and look for Render in the list. On a Team or Enterprise plan, an Owner or Primary Owner has to enable it for the organization before members can connect.
- 02
Start the connection
Click Connect on Render's row, then sign in through the window Render opens itself. If the link breaks later, use Disconnect and run the connection again from the same row.
- 03
Review the authorization screen
Read the authorization screen before you confirm. It belongs to Render, not Claude, and it sets how far the access goes. You can also remove that access from your Render account later.
All 25 tools, grouped by what they do
Render gives Claude 25 tools: 18 that read your account, 7 that change something in it.
Two groups: what Claude reads and what it creates or changes. Tool names stay exactly as Claude shows them.
- 18 read
- 7 write
What Claude reads (18)
18 toolsEighteen tools that look at your Render account, three of which only return a dashboard link.
get_deploy
Pulls up one specific deployment of a service so Claude can tell you about it, without starting anything new. Like every action of this connector, it stays inside the workspace you selected.
get_key_value
Returns the details of one Key Value instance. Nothing in the instance is touched, and Claude only sees what your own Render permissions already let you see.
get_metrics
Fetches performance figures for a Render resource: a service, a Postgres database or a Key Value store. Render warns the answer can come back empty when a metric does not apply to that kind of resource.
get_postgres
Shows the details of one PostgreSQL database hosted on Render. The lookup stays inside the selected workspace and within the permissions of the person who connected the account.
get_selected_workspace
Tells you which Render workspace Claude is operating in right now. Every action of this connector stays inside that workspace, so it is worth checking first.
get_service
Retrieves the details of a single service in your Render workspace. Nothing is created or modified, and the request stays scoped to the workspace selected for the session.
list_deploys
Lists the deployment history of a service. From that list, get_deploy can then open one deployment in detail, and list_logs can pull the matching log lines.
list_key_value
Returns every Key Value instance in your Render account, giving Claude an inventory. To look at one of them in detail, Claude then turns to get_key_value.
list_log_label_values
For a given log label, Claude gets every value it takes across the log lines that match your filters. Handy for deciding what to filter on next.
list_logs
Claude pulls the log lines (the running technical record your services keep) that match the filters you set.
list_postgres_instances
Lists every PostgreSQL database in your Render account so Claude knows what exists. A single database can then be opened with get_postgres or queried, read-only, with query_render_postgres.
list_services
Walks through the services in your Render account and brings back the list. One of them can then be opened in detail with get_service, or checked through its metrics and logs.
list_workspaces
Shows the Render workspaces your account has access to. Render scopes each action of the connector to one selected workspace, so this list is where that choice starts.
query_render_postgres
Runs a read-only SQL query against a Postgres database hosted on Render, then Claude walks you through the result. Render says it tries to keep sensitive values such as connection strings out of the chat, without guaranteeing it.
select_workspace
Sets the workspace Claude works in for the session. Picking one does not add or alter any resource on Render.
update_cron_job
Despite the name, this tool leaves your cron job as it is: according to Render it hands back a link to the job's settings in the dashboard, where you make the adjustment yourself.
update_static_site
Same pattern here: the tool returns the link to the static site's settings page in the Render dashboard. Render does not support direct updates to a static site through the connector, so any adjustment stays manual.
update_web_service
Render states that the connector cannot update a web service directly: Claude gets a link to its settings in the dashboard, and the Render API remains the other route.
What Claude creates or changes (7)
7 toolsSeven tools that add a resource or change a service. No source gives them their own confirmation, so the general rule applies.
create_cron_job
Approval: see the ruleCreates a new cron job in your Render account, meaning a program that runs on a schedule you define.
create_key_value
Approval: see the ruleSpins up a new Key Value instance on Render. It is a fresh resource added to your account.
create_postgres
Approval: see the ruleProvisions a new PostgreSQL database in your Render account.
create_static_site
Approval: see the ruleSets up a new static site in your Render account.
create_web_service
Approval: see the ruleDeploys a new web service in your Render account based on your instructions.
trigger_deploy
Approval: see the ruleStarts a new deployment of a service, with the option to clear the build cache first. Render reserves it for deploys that would not happen automatically.
update_environment_variables
Approval: see the ruleUpdates a service's environment variables, the settings an app reads when it starts. The tool expects the complete list of variables.
What Claude asks before it acts
By default, Claude stops and asks for your go-ahead before each action it takes on an account for you. The request shows up in the chat.
On Team and Enterprise, owners decide whether members can let certain actions through without being asked every time. They can also limit what a connector may do across the whole organization, keeping reads and blocking writes for instance, and nobody can override that from their own account. Claude works with your Render permissions and nothing more. Render's own docs warn that the access you grant covers your workspaces and services, and that some operations can be destructive.
Which plans include the connector
None of the 819 directory sheets shows availability by plan. Connector by connector, that information is not published anywhere.
The general rule is documented: remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile. On Team and Enterprise, an Owner or Primary Owner turns the connector on for the organization before members connect. For the current state, check Render's sheet in the official directory. This connector is also documented by Render alone; Anthropic has no page dedicated to it.
Where the Render connector stops
A connector is not an automation. Claude calls these tools while it answers you: nothing starts by itself when a service crashes or a deploy fails.
Render has deliberately narrowed what a conversation can change. Creating is allowed for five resource types. Editing an existing one is limited to environment variables and triggering a deploy. Deleting is not possible at all; that goes through the Render dashboard or API. The directory badge is not a security audit, and Anthropic notes it does not vouch for how a publisher's tools behave. Connect it knowing exactly what access you hand over.
Need help connecting Render to Claude?
A person reads every message.
