Claude Security integration: share website security findings

Letaido's Claude Security integration lets an AI agent read your scan findings and share them in 76 other apps, on a schedule or when you ask.

Get Letaido

Best Claude Security integration jobs to start with

Website finding recaps, engineering handoffs, a weekly log and a pre-launch check. Open one to watch it run.

Every Monday at 9am UTC

Website security findings in a Slack launch channel

Every Monday the agent posts last week's new critical and high findings for your marketing website to its launch channel in Slack.

Every day at 9am UTC

Website security findings handed to engineering in Linear

Every morning the agent opens a Linear issue for the web engineering team for each new critical or high finding on your marketing website.

Every Monday at 8am UTC

Website open findings logged in Google Sheets

Every Monday the agent adds a row to your sheet with your marketing website's open Claude Security findings by severity.

When you ask

Website security check before a launch

Before a launch, ask the agent to list the open critical and high Claude Security findings on your marketing website, most severe first.

Claude Security integration options, compared

Six ways to work with Claude Security findings, and where each one fits.

Option	Set up by asking	Reaches your team's tools	Runs on its own	No code	Triages findings
Letaido AI agent	Yes	Yes	Yes	Yes	Yes
Claude Security in Claude.ai	No	Partly	Yes	Yes	Yes
Claude Security webhooks	No	Partly	Yes	No	No
Claude Security API	No	Partly	Partly	No	Yes
Claude Security connector (MCP)	Yes	Partly	No	Yes	Yes
Claude Security plugin for Claude Code	Yes	No	No	Partly	No

Read the full comparison

Option	How it works	Best for	Watch out for
Letaido AI agent	You describe the job. The agent reads findings and projects, triages findings and works in 76 other apps, on a schedule or when you ask.	Weekly website finding recaps, Linear issues for engineering and an open-finding log, without code.	77 apps, not thousands. No Claude Security trigger, so jobs run on a schedule or when you ask.
Claude Security in Claude.ai	Projects scan your repositories and list findings with severity, evidence and a suggested fix. You triage them or click Fix to open Claude Code.	Reviewing and fixing findings where Claude finds them.	Reaching other tools takes an export, webhooks or the API.
Claude Security webhooks	Claude Security sends new-finding events to a URL you set up in its settings.	Sending new findings to Slack, Jira or a system you run as they arrive.	Something on your side has to receive each event and act on it.
Claude Security API	Developers use a personal access token to list and read projects and findings, resolve or reopen findings, and manage projects and repositories.	Ticketing, vulnerability management and reporting tools you build.	Needs someone to write and run the code.
Claude Security connector (MCP)	Claude and Claude Code list and read your projects and findings, and resolve or reopen a finding when you ask.	An engineer pulling up a finding, fixing it and resolving it in one session.	Works while you chat. It doesn't run jobs on its own.
Claude Security plugin for Claude Code	The /claude-security command scans a repository or a set of changes in your Claude Code session and writes a report and suggested patches to a folder in your repository.	Checking a branch, a pull request or a commit before you merge.	Runs while Claude Code stays open. You apply each patch yourself.

How the Claude Security integration works

Connect Claude Security once. After that, you describe each job in plain English.

1Paste a Claude Security API key
2Ask for a job in a chat, or start from an example above
3Approve the first run
Allow all read operations once per app; approve each kind of write once.
4Watch the first run in the chat
Save it to run on a schedule.

How a Claude Security job starts

A schedule starts the job, or you do by asking.

On a scheduleEvery morning or every Monday, looking back over a fixed window like the past day or last week, in UTC.
When you askType the job in a chat, like listing the website's open findings before a launch, and it runs once.

Setup details

Turn on Claude Security

It's in beta on Claude Enterprise plans. An owner turns it on in Organization settings, Claude Security.

Create a personal access token

In Claude Security settings, create a token. It acts as you, so the agent sees only the projects you can see.

Paste it into Letaido

Add the token. If it isn't limited to one workspace, add the workspace id too.

Watch the first run

Allow all read operations once per app; approve each kind of write once. Then save it to repeat.

Every Claude Security action the agent can take

15 actions across findings, projects and the repositories they scan.

ReadsWrites

Related integrations

Frequently asked

How do I connect Claude Security to Letaido? +

Create a personal access token in Claude Security settings and paste it into Letaido. Add the workspace id if the token isn't limited to one workspace. The token acts as you, so Leta sees only the projects you can see.

What do I need before I connect? +

Claude Security, which is in beta on Claude Enterprise plans and turned on by an owner. To add repositories through Letaido, the Claude GitHub App must reach them and your GitHub account must be connected to Claude.

Can a new finding start a job? +

No. Claude Security sends its new-finding webhooks to systems you run, not to Letaido. Jobs run on a schedule, like the daily Linear issues, or when you ask.

Will Letaido start scans that cost money? +

Reading and triaging findings doesn't start scans. Creating a project, adding or changing a repository and unarchiving a project do, and they add to your Anthropic organization's costs. Leta says what will be scanned before it does any of these.

Is there a Claude Security MCP server? +

Yes. The Claude Security connector lets Claude and Claude Code list and read projects and findings and resolve or reopen a finding when you ask. Letaido is not an MCP server. It has its own Claude Security connector, uses it with 76 other apps and runs jobs on a schedule or when you ask.

How is Letaido different from the Claude Security plugin for Claude Code? +

The plugin scans a repository or a set of changes in your Claude Code session and writes a report and suggested patches for you to apply. Letaido doesn't scan code itself. It reads the findings your Claude Security projects already have and gets them to tools like Slack, Linear and Google Sheets. Adding a repository through Letaido makes Claude Security scan it.

Which actions need my approval? +

Allow all read operations once per app; approve each kind of write once.

Get your website's security findings to the right people

Connect Claude Security, then ask for a job like "Post last week's critical and high findings for our website to #website-launch every Monday". Leta gets it done on your schedule, or the moment you ask.