Snyk Studio MCP Server: Setup for Claude, Cursor and Other Apps

• Checks your own source code for security flaws

  Runs a static code scan on a folder and lists the security weaknesses it finds, without running your code.

  Try: “Scan this project's source code for security issues.”

  Files · snyk_code_scan

• Finds known vulnerable open-source packages in a project

  Reads manifest files such as package.json or pom.xml, checks the dependencies against Snyk's vulnerability database and flags license problems. For some ecosystems it may run build tools such as Gradle or Maven to list the dependencies.

  Try: “Check this project's dependencies for known vulnerabilities.”

  Runs code · snyk_sca_scan

• Looks for passwords and API keys left in your code

  Scans source and configuration files for hardcoded secrets such as keys, tokens and passwords.

  Try: “Did I leave any API keys in this repo?”

  Files · snyk_secret_scan

• Checks infrastructure files for risky settings

  Analyzes Terraform, Kubernetes, CloudFormation, Azure ARM and Serverless files for security misconfigurations.

  Try: “Scan the Terraform in the infra folder for misconfigurations.”

  Files · snyk_iac_scan

• Checks a container image for known vulnerabilities

  Scans an image for flaws in its operating system packages and app dependencies. You can pass a Dockerfile for extra context.

  Try: “Scan the image my-api:v2 and show the high-severity issues.”

  Network · snyk_container_scan

• Checks a software bill of materials file for vulnerabilities

  Tests the components listed in an existing SBOM file against Snyk's vulnerability data. Components must be identified with package URLs.

  Try: “Test this SBOM file for known vulnerabilities.”

  Files · snyk_sbom_scan

• Rates a package's security and upkeep before you add it

  Returns security issues, maintenance status, popularity and community health for a package and version.

  Try: “Is the left-pad package healthy enough to add as a dependency?”

  Network · snyk_package_health_check

• Estimates whether a package upgrade will break your code

  Runs a breaking change assessment for upgrading a package from one version to another.

  Try: “Will upgrading lodash from 4.17.15 to 4.17.21 break anything?”

  Network · snyk_breakability_check

• Lists the AI models and datasets a Python project uses

  Generates an AI bill of materials in CycloneDX format for a local Python project. It needs an internet connection, the experimental feature (available from Snyk on request) and the preview release of the CLI.

  Try: “Create an AI bill of materials for this Python project.”

  Write · snyk_aibom

• Signs the assistant in to your Snyk account

  Starts the Snyk sign-in when a scan reports that you are not authenticated.

  Try: “Sign me in to Snyk.”

  Network · snyk_auth

• Signs you out of Snyk on this computer

  Clears the Snyk token stored on your machine, useful when you switch accounts.

  Try: “Log me out of Snyk.”

  Write · snyk_logout

• Marks a folder as safe for Snyk to scan

  Trusts a given folder so scans may run in it. The tool says to run it only when told to.

  Try: “Trust this folder so Snyk can scan it.”

  Write · snyk_trust

• Shows which Snyk version is installed

  Returns the installed Snyk MCP version, handy when reporting a problem.

  Try: “Which version of the Snyk server am I running?”

  Read · snyk_version

• Tells Snyk how many issues were prevented or fixed

  Reports counts of security issues the assistant prevented or fixed in this run, with severity breakdowns.

  Try: “Report that you fixed two high-severity issues.”

  Write · snyk_send_feedback

Watch out for

• Dependency scans may run build tools

  The open-source scan can run tools such as Gradle or Maven on your machine to read a project's dependency tree. Only scan projects you trust. github.com

• It can mark folders as trusted

  A trust tool lets the assistant approve a folder for scanning. Its description says to run it only when told to, so watch for it in the tool-call approvals. github.com

• The lite profile is not a read-only mode

  Profiles only reduce the tool list. Even the lite profile includes the trust, sign-out and feedback tools. github.com

Security advisories