For AI agents: A markdown version of this page is available at https://docs.datadoghq.com/security/code_security/dev_tool_int/mcp_server/tools_reference.md. A documentation index is available at /llms.txt.
This product is not supported for your selected Datadog site. ().

The Datadog Code Security MCP Server exposes the following tools for AI coding assistants and CLI usage. Scan tools wrap one or more Datadog security binaries and accept file paths or directories to scan. datadog_library_vulnerability_scan queries Datadog by package URL and does not scan a local tree.

This MCP server is separate from the Datadog MCP Server, which provides cloud-based access to Datadog features and data. The Code Security MCP Server runs locally and focuses on code-level security scanning. To query findings already stored in Datadog, see Query findings already in Datadog.

datadog_code_security_scan

Run SAST, secrets detection, SCA, and IaC scanning in parallel.

min_sast_severity applies only to the SAST portion of the scan. It defaults to LOW. In-source suppressed findings are excluded. When some scanners are not installed, the tool returns results from the scanners that succeeded and an error for each scanner that failed.

Parameters

ParameterTypeRequiredDescription
file_pathsarray[string]YesFile paths or directories to scan
working_dirstringNoBase directory for resolving relative paths (defaults to the current directory)
min_sast_severitystringNoMinimum SAST severity to return: LOW, MEDIUM, HIGH, or CRITICAL. Default: LOW. SAST only.

Required binaries

datadog-static-analyzer, datadog-sbom-generator, datadog-security-cli, datadog-iac-scanner

datadog_sast_scan

Run Static Application Security Testing (SAST) to detect security vulnerabilities in first-party code.

min_sast_severity defaults to LOW. In-source suppressed findings are excluded.

Parameters

ParameterTypeRequiredDescription
file_pathsarray[string]YesFile paths or directories to scan
working_dirstringNoBase directory for resolving relative paths
min_sast_severitystringNoMinimum SAST severity to return: LOW, MEDIUM, HIGH, or CRITICAL. Default: LOW.

Required binary

datadog-static-analyzer

datadog_secrets_scan

Detect hardcoded credentials, API keys, passwords, and tokens in source code and configuration files.

Parameters

ParameterTypeRequiredDescription
file_pathsarray[string]YesFile paths or directories to scan
working_dirstringNoBase directory for resolving relative paths

Required binary

datadog-static-analyzer

datadog_sca_scan

Run SCA to detect known vulnerabilities (CVEs) in your project’s dependencies. This tool performs a two-step process:

  1. Generates a Software Bill of Materials (SBOM) from the specified directories.
  2. Scans the SBOM for known vulnerabilities using Datadog’s vulnerability database.

Parameters

ParameterTypeRequiredDescription
file_pathsarray[string]YesDirectories to scan for dependencies
working_dirstringNoBase directory for resolving relative paths

Output

Vulnerabilities with CVE ID, severity, affected component, version, and description.

Required binaries

datadog-sbom-generator, datadog-security-cli

datadog_iac_scan

Detect misconfigurations, compliance issues, and security vulnerabilities in IaC files.

Parameters

ParameterTypeRequiredDescription
file_pathsarray[string]YesDirectories containing IaC files to scan
working_dirstringNoBase directory for resolving relative paths

Output

Security findings with severity, rule, file location, and remediation guidance.

Required binary

datadog-iac-scanner

Supported IaC formats

  • Terraform
  • CloudFormation
  • Kubernetes manifests
  • Dockerfiles
  • GitHub Actions

datadog_generate_sbom

Generate a comprehensive SBOM listing all software components, dependencies, versions, and licenses in a repository.

Parameters

ParameterTypeRequiredDescription
pathstringNoPath to repository or directory to analyze (defaults to current directory)
working_dirstringNoBase directory for the scan (defaults to current directory)

Output

JSON containing a summary (total components, breakdown by language/package manager, license statistics) and a detailed component list (name, version, type, license, package URL).

Supported package managers

LanguagePackage managers
.NETNuGet
C++Conan
GoGo modules
JavaGradle, Maven
JavaScriptnpm, pnpm, Yarn
PHPComposer
Pythonpdm, pipenv, poetry, requirements.txt, uv
RubyBundler
RustCargo
If the repository uses a package manager not listed above, or if the tool returns zero components, the AI assistant can generate an SBOM by reading lock files (package.json, requirements.txt, go.mod, pom.xml, Gemfile.lock, Cargo.lock, composer.lock, and similar files) and extracting dependencies directly.

datadog_library_vulnerability_scan

Look up known vulnerabilities for specific libraries by package URL (PURL). This tool calls the Datadog API and does not scan a local codebase. Use datadog_sca_scan when you want to scan a project tree.

Requires DD_API_KEY and DD_APP_KEY.

Parameters

ParameterTypeRequiredDescription
librariesarray[object]YesLibraries to scan. Each object uses the fields in the table below.
working_dirstringNoWorking directory for Git context detection (defaults to the current directory)

Each entry in libraries:

FieldTypeRequiredDescription
purlstringYesPackage URL, for example pkg:maven/com.cronutils/cron-utils@9.1.2
is_directbooleanNoWhether this is a direct dependency
is_devbooleanNoWhether this is a development-only dependency
package_managerstringNoPackage manager, for example MAVEN, NPM, or GOLANG

Output

Vulnerabilities with CVE ID, GHSA ID, severity, CVSS score, affected library, remediation, fix versions, and exploit availability.

datadog_ephemeral_sast_scan

Run one local, unpublished SAST rule against a repository. The caller supplies the rule JavaScript and the tree-sitter query. The tool does not require DD_API_KEY or DD_APP_KEY.

The scan runs only the supplied rule but still follows the repository’s Static Code Analysis configuration: ignore-paths, only-paths, max-file-size-kb, and use-gitignore.

Parameters

ParameterTypeRequiredDescription
rule_codestringYesRule JavaScript source, not base64-encoded
tree_sitter_querystringYesTree-sitter query source, not base64-encoded
languagestringYesLanguage of the rule, for example TYPESCRIPT, JAVASCRIPT, PYTHON, GO, or JAVA
directorystringNoRepository directory to scan (defaults to the current directory)
rule_namestringNoRule name inside the temporary ruleset
pathsarray[string]NoFiles or subdirectories inside directory to scan. Omit to scan the whole directory.

Output

Violations reported by the rule, grouped by file. If the rule throws an exception, fails to compile, or times out on any file, the tool returns an error that lists those files instead.

Required binary

datadog-static-analyzer version 0.9.10 or later. If you have an earlier version, see Scanner upgrade required.

datadog_promote_sast_rule

Save an approved SAST rule as a custom rule in Datadog. The tool uploads the rule exactly as it receives it, and only after the user approves it. Before saving, the tool runs the rule on the two tests examples and saves nothing if either example’s finding count differs from its annotation_count or if the rule fails. If a rule with the same name exists, the tool does not replace it unless overwrite is true.

Scans use the saved rule only after the repository configuration lists its ruleset (security-variants by default). Add it under sast.use-rulesets in code-security.datadog.yaml. When the only configuration file is static-analysis.datadog.yml or static-analysis.datadog.yaml, add the ruleset to its rulesets list. If both files exist, code-security.datadog.yaml takes precedence over the legacy file. See Static Code Analysis configuration.

Requires DD_API_KEY and DD_APP_KEY.

Parameters

ParameterTypeRequiredDescription
user_confirmedbooleanYesMust be true. Set it only after the user approves the rule.
rule_codestringYesRule JavaScript source, not base64-encoded
tree_sitter_querystringYesTree-sitter query source, not base64-encoded
languagestringYesLanguage of the rule, for example TYPESCRIPT
rule_namestringYesLowercase rule name, 5-32 characters
ruleset_namestringNoLowercase ruleset name, 5-32 characters. Default: security-variants.
overwritebooleanNoReplace a rule that already exists. Default: false.
short_descriptionstringNoOne-line summary, at most 60 characters
descriptionstringNoDescription stored on the rule
testsarray[object]YesExactly two examples: the fixed code and the code before the fix. Each object uses the fields in the table below.

Rule and ruleset names can contain lowercase letters, digits, and single hyphens. They cannot start or end with a hyphen.

Each entry in tests:

FieldTypeRequiredDescription
filenamestringYesFilename without directories. Its extension must match the rule’s language, for example .ts for TYPESCRIPT.
codestringYesExample source, not base64-encoded
annotation_countintegerYesNumber of findings the rule must report on this example: 0 for the fixed code, and a positive number (usually 1) for the code before the fix

Output

The ruleset name, rule name, and revision ID of the saved rule.

Required binary

datadog-static-analyzer version 0.9.10 or later, to run the rule on the examples before saving.

Further reading