Skip to content
For the complete documentation index, see llms.txt. This page is also available as Markdown.

Traceable MCP Server

Connect AI assistants and agents to Traceable using the Traceable MCP Server to query API inventory, findings, and runtime behavior

Updates (October 2026 to December 2026)
  • October 2026: Added support for Self-Managed Platform (SMP) deployments. For more information about connecting to the MCP server on SMP, see Configure the MCP server.

Older

  • August 2026: Updated the page to add information about the availability of AI asset-related data in tools. Go to Available tools and capabilities for more details.

The Traceable MCP (Model Context Protocol) Server enables you, AI assistants, and agents to securely interact with Traceable using MCP. It exposes Traceable’s API security data, observations, and analysis capabilities in a structured, machine-consumable format, while enforcing the same authentication and authorization controls as the Traceable platform.

By integrating with MCP, Traceable allows AI systems to query API inventory, analyze security findings, and understand runtime behavior without direct access to raw infrastructure or internal systems.

Deployment support

The Traceable MCP server is available for both SaaS and Self-Managed Platform (SMP) deployments. The configuration steps are the same for both. The only difference is the endpoint URL you connect to:

  • SaaS: The MCP server is hosted by Traceable and reached through your Traceable cluster’s public URL, for example, https://api.traceable.ai/mcp.

  • SMP: The MCP server runs in your own environment and is reached through your SMP tenant’s domain, https://api.<CUSTOMER_TENANT_FQDN>/mcp. Depending on your network setup, this URL may be reachable only through your VPN.

For more information, see Public MCP endpoint.

What you will learn from this topic

By the end of this topic, you will understand:

  • What the Traceable MCP server is and how it works.

  • The prerequisites required to configure and access the MCP server.

  • The tools and capabilities available through MCP.

  • The data sources through which Traceable extracts data.

  • The security considerations and limitations when using the MCP.

  • Why and when you should use the MCP server.

  • How Traceable handles the authentication for the MCP server.

  • The steps to configure the MCP server.


Before you begin

Make a note of the following before configuring your MCP server:

  • Make sure that you have access to the Traceable Public URL for your SaaS cluster or SMP tenant. For more information about the URL format for each deployment, see Public MCP endpoint.

  • Make sure that you have a valid Platform API token. For the steps to retrieve this, go to Public APIs. This token is required to authenticate the user and access the MCP server.

  • Make sure you are connected to the VPN if your SMP environment is accessible only through one.


Available tools and capabilities

The Traceable MCP server exposes the following set of tools that allow AI agents and clients to query and analyze API security data:

Tool Category
Tool
Description

Query and Data Retrieval

get_attribute_mappings

Returns valid attribute mappings for entities like API, SERVICE, DOMAIN, BACKEND, ACTOR, and BACKEND_API.

You must call this tool before calling get_graphql_query_template_details.

get_graphql_query_template_details

Returns GraphQL query templates for supported queries, such as entities, vulnerabilitiesV3, explore, searchAst, threatActivity, and submitAnalyticsQueryBatch. Go to Supported query types for more details. You must call this before executing any GraphQL query.

execute_graphql_query

Executes a valid GraphQL query against the Traceable platform.

You must call the above tool first.

lookup_entity_details

Looks up entity details using an entity name or ID.

API Specific

find_apis_by_regex

Searched for APIs using a regex pattern.

get_api_insights_for_api_id

Retrieves the OpenAPI specifications for a specified API ID.

get_sample_spans_for_api_id

Retrieves sample trace spans for a specified API ID.

Time and Filtering

get_query_time_range

Returns UTC start and end timestamps for time-based filtering. Accepts duration, start time, or end time (in milliseconds).

get_threat_activity_timerange

Generates optimized time ranges for threat activity queries, expanding windows to capture boundary-spanning events.

get_vuln_enum_filter_supported_values

Returns supported enum values for vulnerability filters, such as STATUS, SEVERITY, OWASP, and VULNERABILITY_CATEGORY.

Sensitive Data

get_datatypes_details_from_datatype_names

Retrieves metadata for sensitive data types (ID, sensitivity level, suppression status) using data type names such as Personal Identification or Financial Information.

get_datatypes_details_from_datatype_ids

Retrieves sensitive data type metadata using dataTypeId values obtained from prior queries or entity metadata.

Trends and Display

calculate_api_count_n_days_ago

Calculates the historical count of APIs as of a specified number of days in the past.

get_display_urls

Generates clickable display URLs for entities, threat activities, or explore results. Typically used at the end of workflows.

Supported query types

The get_graphql_query_template_details tool supports the following query types:

Query Name
Description

entities

Fetches entities, such as API, MCP, SERVICE, DOMAIN, BACKEND, ACTOR, and BACKEND_API.

vulnerabilitiesV3

Fetches issues (vulnerabilities) from runtime, gateway, compliance, security testing, and conformance sources.

explore

Fetches explore-based datasets, for example, DOMAIN_EVENT.

searchAst

Fetches API testing scan data across multiple scans.

threatActivityMetadata

Fetches metadata for threat activity fields.

threatActivity

Fetches threat activity data.

submitAnalyticsQueryBatch

Fetches API risk, inventory, updates, deletions, and time-based vulnerability trends.

getAstHookTests

Fetches API testing authentication hook test details.

mcpToolInsightByMcpToolId

Fetches insight data for a specific MCP tool using its mcpToolId.

mcpPromptInsightByMcpPromptId

Fetches insight data for a specific MCP prompt using its mcpPromptId.

mcpResourceInsightByMcpResourceId

Fetches insight data for a specific MCP resource using its mcpResourceId.


Data sources

The MCP server enables you to query and extract data from the following modules and their corresponding sources:

Module
Sources

Discovery

  • Inventory: Provides a detailed catalog of all discovered APIs and AI/MCP assets (MCP tools, MCP servers, MCP prompts, and MCP resources), including their endpoints, authentication requirements, exposure levels (internal or external), ownership information, and traffic visibility. Go to Inventory for more details.

  • Services: Provides a logical grouping of APIs that help you understand which APIs belong to which service and the associated risk. Go to Services for more details.

  • Domains: Provides information on internal and external domains connected to the environments, highlighting which domains are public-facing, internal-only, or linked to third parties. Go to Domains for more details.

  • Backends: Provides information on connected backend systems and services that receive traffic from APIs, allowing you to identify dependencies, traffic volume, and data sensitivity within backend connections. Go to Backends for more details.

  • Sensitive Data: Provides information on API endpoints that process sensitive data, such as Personally Identifiable Information (PII), Social Security Numbers (SSN), or authentication tokens. This helps you identify risks associated with data privacy. Go to Sensitive Data for more details.

  • Posture Events: Provides information on discovery-related changes in your API ecosystem. It includes events such as the discovery of issues and the introduction of new services or APIs. This helps you track changes in your security posture over time and highlights areas that may require investigation or remediation. Go to Posture Events for more details.

Protection

  • Threat Actors: Provides information about malicious sources or entities identified by Traceable as potential attackers. This includes details, such as attacker information, origin, activity patterns, and severity levels. This helps you understand who is targeting your APIs. Go to Threat Actors for more details.

  • Threat Activities: Provides information about detected attack activities, including APIs and users. This provides visibility into attempted exploits, attack vectors, timestamps, and affected APIs, helping you track threat behavior. Go to Threat Activity for more details.

  • Security Events: Provides information on all security-related events captured across your environment. It displays data, such as detections, anomalies, and rules triggered by the threat actor, allowing you to analyze security incidents within your application.

  • APIs under Threat: Provides information on the APIs being targeted or showing signs of attack. It provides insights into affected APIs, associated threat types, attack frequency, and risk levels to help prioritize protection measures. Go to APIs under Threat for more details.


Security considerations

The Traceable MCP server is created with security as the core requirement. When you use the MCP server:

  • All requests are authenticated using your Traceable Platform API token.

  • Access to data is enforced using the same role-based access control that applies across the Traceable platform.

  • You or the AI agent can access only the data permitted for your assigned roles.

  • The prompts you use determine whether Traceable returns IP addresses or sensitive information in the results.

By default, the MCP Server is created with read-only access, helping reduce the risk of unintended changes to your security posture.


Limitations

Make note of the following limitations when using the Traceable MCP server:

  • Policy creation, modification, or enforcement is not available through the MCP server.

  • Access to raw request or response payloads may be restricted based on your application configuration in the Traceable platform.

  • MCP capabilities are limited to what the authenticated user is permitted to access in the Traceable platform.


Use-cases of the Traceable MCP server

The Traceable MCP Server is built for you to use AI assistants or agents with your API security data.

You should use Traceable MCP when you want to:

  • Allow AI assistants to query API inventory and security findings using prompts.

  • Investigate API threats and unusual behavior faster using natural language queries.

  • Provide AI tools with controlled, read-only access to API security data.

  • Understand how runtime API behavior maps to security policies without manually navigating the Traceable platform.

The MCP Server ensures that AI access to Traceable data is secure, limited to the appropriate scope, and auditable, making it suitable for production environments.


Configure the MCP server

Traceable provides its MCP server, which you can configure to securely connect AI assistants or agents to your Traceable environment.

Public MCP endpoint

Traceable exposes the MCP server through the public URL associated with your Traceable cluster. You can specify this URL while configuring the MCP server:

SaaS:

SMP:

Authentication

The Traceable MCP server uses token-based authentication. This token defines the tenant you can access. For more information about generating and copying a Platform API token, go to Public APIs.

Once you generate the token, specify it in the Authorization header using the following code block while configuring the MCP server:

Configuration

Traceable provides multiple ways of configuring the MCP server depending on the coding assistant you use:

To configure the Traceable MCP server, navigate to your Claude CLI and execute the following command:

To configure the Traceable MCP server, navigate to your Claude Desktop app and complete the following steps:

  1. Navigate to the app's Settings page.

  2. Under Desktop app, click Developer.

  3. In the MCP servers page, click Edit Config. Claude redirects you to the claude_desktop_config.json file.

  4. Open the file using an editor of your choice and specify the following code:

  5. Save the file.

  6. Quit the Claude Desktop app, then restart it.

You can now use the Claude Desktop app to interact with the Traceable MCP server.

To configure the Traceable MCP server, navigate to your Cursor app and execute the following command in the mcp.json file. Go to Installing MCP servers for more details.

To configure the Traceable MCP server, navigate to your Windsurf app and execute the following command in the mcp.json file. Go to Connecting to MCP servers for more details.

To configure the Traceable MCP server, navigate to your VS Code app and execute the following command in the mcp.json file. Go to Add an MCP server for more details.

Last updated

Was this helpful?