The Postman MCP Server connects AI tools to Postman's API platform, enabling AI agents and assistants to manage Postman resources, automate workflows, and interact using natural language.
Capabilities:
Collection Management: Create, update, duplicate, and retrieve collections; add requests and responses; run collections with environment variables and iteration control; sync bidirectionally with API specifications.
Workspace Management: Create and manage workspaces with different visibility types (personal, private, public, team, partner); list and filter workspaces by type and creator.
Environment Management: Create, update, and retrieve environments; manage environment variables including secret and default types with descriptions.
API Specification Management: Create and manage single or multi-file specifications (OpenAPI 3.0, AsyncAPI 2.0); generate collections from specs and specs from collections; sync specifications and collections bidirectionally.
Mock Server Management: Create, update, publish, and retrieve mock servers linked to collections.
Code Generation: Generate production-ready client code from API definitions (available in "Code" tool configuration).
Discovery & Search: Get authenticated user information; retrieve entities by tags; check status of asynchronous tasks; discover enabled tools.
Advanced Features: Support for multiple authentication types (Basic, Bearer, OAuth1/2, AWS Signature, JWT, API Key, and more); configure protocol profiles (SSL, redirects, HTTP versions, TLS); handle async operations with polling.
Multiple Tool Configurations: Operate in Minimal (37 essential tools), Full (100+ tools), or Code (code generation) modes to suit different use cases.
Provides comprehensive tools for managing Postman workspaces, collections, environments, and API requests. Enables programmatic interaction with the Postman API including creating, updating, and deleting collections, managing environments, and performing API operations.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Postman run the latest test suite from my 'Payment API' collection"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
The Postman MCP Server implements the Model Context Protocol (MCP) to connect AI agents and coding assistants — including Claude Code, Cursor, VS Code Copilot, GitHub Copilot CLI, and Gemini CLI — directly to your Postman workspaces, collections, specifications, and environments.
Postman also offers the server as an npm package.
For the full installation guide with agent-specific setup, see Postman's MCP Server product page.
The Postman MCP Server collection is the quickest way to explore, test, and connect to the Postman MCP Server. Use it to:
Browse the complete list of available tools across all configurations.
Connect to and test the local server.
Related MCP server: Postman MCP Generator
Minimal — (Default) Only includes essential tools for basic Postman operations. Ideal for users who want to modify a single Postman element, such as collections, workspaces, or environments.
Code — Includes tools to generate high-quality, well-organized client code from public and internal API definitions. Ideal for users who need to consume APIs or get API context to their agents.
Full — Includes all available Postman API tools (100+ tools). Ideal for users who engage in advanced collaboration and Postman's Enterprise features.
Learn — Searches Postman Docs for guides, tutorials, and reference content. Ideal for agents who need to discover Postman features, look up API concepts, or find learning resources.
For the best developer experience and fastest setup, use OAuth on the remote server (https://mcp.postman.com). OAuth is fully compliant with the MCP Authorization specification and requires no manual API key configuration.
The EU remote server and the local server support only Postman API key authentication.
Remote (any OAuth-compatible MCP host):
Add this URL to your MCP host's configuration:
https://mcp.postman.com/minimalChange /minimal to /code or /mcp for Code or Full mode. For EU or API key auth, pass Authorization: Bearer <POSTMAN_API_KEY> as a header.
Local:
npx @postman/postman-mcp-serverAdd --code or --full for Code or Full mode. Set POSTMAN_API_KEY as an environment variable.
For IDE-specific setup instructions, see the following table. For more information, see the Postman MCP Server docs.
Agent / IDE | Remote | Local |
Claude Code | ||
Claude Desktop | ||
Cursor | ||
VS Code | ||
Codex | ||
Antigravity CLI | ||
GitHub Copilot CLI | ||
Kiro | ||
Docker | — |
The Postman MCP Server supports the EU region for remote and local servers:
For streamable HTTP, the remote server is available at https://mcp.eu.postman.com/mcp (Full), https://mcp.eu.postman.com/code, and https://mcp.eu.postman.com/minimal.
For the STDIO public package, use the --region eu flag, or set the POSTMAN_API_BASE_URL environment variable directly.
OAuth isn't supported for the EU server. The EU remote server only supports API key authentication.
API Testing — Continuously test your API using your Postman collection. Use the local server to test local APIs, as the remote server won't have network access to your workstation.
Code synchronization — Keep your code in sync with your Postman Collections and specs.
Collection management — Create and tag collections, update documentation, add comments, or perform actions across multiple collections without leaving your editor.
Workspace and environment management — Create workspaces and environments, plus manage environment variables.
Automatic spec creation — Create specs from your code and use them to generate collections.
Client code generation — Generate production-ready client code that consumes APIs following best practices and project conventions.
For Docker setup and installation, see DOCKER.md.
Bug reports, tool requests, and documentation fixes are all welcome — see CONTRIBUTING.md.
The MCP tool definitions and the server implementation here are synced from Postman's
internal source of truth, so changes to files under src/ can't be merged in this
repository — but issues are how those changes get made, and we credit contributions
that ship. Documentation and repository tooling accept pull requests directly.
Found a security issue? See SECURITY.md — please don't open a public issue.
See Add your MCP requests to your collections to learn how to use Postman to perform MCP requests.
Visit the Postman Community to share what you've built, ask questions, and get help.
You can connect to both the remote and local servers and test them using the Postman MCP Server collection.
Report bugs and request tools in GitHub Issues. See SUPPORT.md for which channel fits your question.
Model Context Protocol · MCP Server · Postman · AI Agents · Claude Code · Cursor · VS Code · Specifications · REST API · API Testing · TypeScript · OpenAPI
Creates a collection using the Postman Collection v2.1.0 schema format.
Note:
If you do not include the `workspace` query parameter, the system creates the collection in the oldest personal Internal workspace you own.
Inputs: workspace (required), collection.
Creates a request in a collection. For a complete list of properties, refer to the Request entry in the Postman Collection Format documentation.
Note:
It is recommended that you pass the `name` property in the request body. If you do not, the system uses a null value. As a result, this creates a request with a blank name.
Inputs: collectionId (required), url, auth, data, +11 more.
Creates a request response in a collection. For a complete list of request body properties, refer to the Response entry in the Postman Collection Format documentation.
Note:
It is recommended that you pass the `name` property in the request body. If you do not, the system uses a null value. As a result, this creates a response with a blank name.
Inputs: request (required), collectionId (required), url, mime, +15 more.
Creates an environment.
Note:
The request body size cannot exceed the maximum allowed size of 30MB.
If you receive an HTTP `411 Length Required` error response, manually pass the `Content-Length` header and its value in the request header.
If you do not include the `workspace` query parameter, the system creates the environment in the oldest personal Internal workspace you own.
Only shared variable values can be modified through the Postman API. A shared variable is an environment variable with its value synced and stored in the Postman cloud, and can be accessed by your teammates in the environment's workspace.
Inputs: workspace (required), environment.
Creates a mock server in a collection.
Pass the collection UID (ownerId-collectionId), not the bare collection ID.
If you only have a `collectionId`, resolve the UID first:
Prefer GET `/collections/{collectionId}` and read `uid`, or
Construct `{ownerId}-{collectionId}` using ownerId from GET `/me`:
For team-owned collections: `ownerId = me.teamId`
For personal collections: `ownerId = me.user.id`
Use the `workspace` query to place the mock in a specific workspace. Prefer explicit workspace scoping.
Inputs: workspace (required), mock.
Creates an API specification in Postman's Spec Hub. Specifications can be single or multi-file.
Note:
Postman supports OpenAPI (2.0, 3.0, and 3.1), AsyncAPI (2.0 and 3.0), protobuf (2 and 3), GraphQL, and Smithy specifications.
If the file path contains a `/` (forward slash) character, then a folder is created. For example, if the path is the `components/schemas.json` value, then a `components` folder is created with the `schemas.json` file inside.
Multi-file specifications can only have one root file.
Files cannot exceed a maximum of 12 MB in size.
Inputs: name (required), type (required), files (required), workspaceId (required).
Creates a file for an OpenAPI or a protobuf 2 or 3 specification.
Note:
If the file path contains a `/` (forward slash) character, then a folder is created. For example, if the path is the `components/schemas.json` value, then a `components` folder is created with the `schemas.json` file inside.
Creating a spec file assigns it the `DEFAULT` file type.
Multi-file specifications can only have one root file.
Files cannot exceed a maximum of 10 MB in size.
Inputs: path (required), specId (required), content (required).
Creates a new workspace.
Note:
This endpoint returns a 403 `Forbidden` response if the user does not have permission to create workspaces. Admins and Super Admins can configure workspace permissions to restrict users and/or user groups from creating workspaces or require approvals for the creation of team workspaces.
Private and Partner Workspaces are available on Postman Team and Enterprise plans.
There are rate limits when publishing public workspaces.
Public team workspace names must be unique.
The `teamId` property must be passed in the request body if Postman Organizations is enabled.
Inputs: workspace.
Creates a duplicate of the given collection in another workspace.
Use the GET `/collection-duplicate-tasks/{taskId}` endpoint to get the duplication task's current status.
Inputs: workspace (required), collectionId (required), suffix.
Creates a collection from the given API specification. The specification must already exist or be created before it can be used to generate a collection. The response contains a polling link to the task status.
Inputs: name (required), specId (required), options (required), elementType (required).
Generates an OpenAPI 2.0, 3.0, or 3.1 specification for the given collection. The response contains a polling link to the task status.
Inputs: name (required), type (required), format (required), elementType (required), +1 more.
Gets all API specifications in a workspace.
Inputs: workspaceId (required), limit, cursor.
Gets information about the authenticated user.
This endpoint provides “current user” context (`user.id`, `username`, `teamId`, roles).
When a user asks for “my …” (e.g., “my workspaces, my information, etc.”), call this first to resolve the user ID.
Inputs: no named parameters.
Get information about a collection. By default this tool returns the lightweight collection map (metadata + recursive itemRefs). Use the model parameter to opt in to Postman's full API responses:
model=minimal — root-level folder/request IDs only
model=full — full Postman collection payload.
Inputs: collectionId (required), model, access_key.
The workspace ID query is required for this endpoint. If not provided, the LLM should ask the user to provide it.
Inputs: workspace (required), name, limit, offset.
Gets the status of a collection duplication task.
Inputs: taskId (required).
IMPORTANT: Run this tool first when a requested tool is unavailable. Returns information about which tools are enabled in the full and minimal tool sets, helping you identify available alternatives.
Inputs: no named parameters.
Gets information about an environment.
Inputs: environmentId (required).
Gets information about all of your environments.
Inputs: workspace.
Gets the API specification generated for the given collection.
Inputs: elementType (required), collectionUid (required).
Gets information about a mock server.
Resource: Mock server entity. Response includes the associated `collection` UID and `mockUrl`.
Use the `collection` UID to navigate back to the source collection.
Inputs: mockId (required).
Gets all active mock servers. By default, returns only mock servers you created across all workspaces.
Always pass either the `workspace` or `teamId` query to scope results. Prefer `workspace` when known.
If you need team-scoped results, set `teamId` from the current user: call GET `/me` and use `me.teamId`.
If both `teamId` and `workspace` are passed, only `workspace` is used.
Inputs: teamId, workspace.
Gets information about an API specification.
Inputs: specId (required).
Gets all of an API specification's generated collections.
Inputs: specId (required), elementType (required), limit, cursor.
Gets the complete contents of an OpenAPI or AsyncAPI specification's definition.
Inputs: specId (required).
Gets the contents of an API specification's file.
Inputs: specId (required), filePath (required).
Gets all the files in an API specification.
Inputs: specId (required).
Requires an Enterprise plan. Tagging is only available on Postman Enterprise plans. This tool returns a 404 error on Free, Basic, and Professional accounts.
Gets Postman elements (entities) by a given tag. Tags enable you to organize and search workspaces, APIs, and collections that contain shared tags.
Inputs: slug (required), limit, cursor, direction, +1 more.
Gets information about a workspace.
Note:
This endpoint's response contains the `visibility` field. Visibility determines who can access the workspace:
`personal` — Only you can access the workspace.
`team` — All team members can access the workspace.
`private` — Only invited team members can access the workspace (Team and Enterprise plans only).
`public` — Everyone can access the workspace.
`partner` — Only invited team members and partners can access the workspace (Team and Enterprise plans only).
Inputs: workspaceId (required), include.
Gets all workspaces you have access to.
For “my …” requests, first call GET `/me` and pass `createdBy={me.user.id}`.
This endpoint's response contains the visibility field. Visibility determines who can access the workspace:
`personal` — Only you can access the workspace.
`team` — All team members can access the workspace.
`private` — Only invited team members can access the workspace (Professional and Enterprise).
`public` — Everyone can access the workspace.
`partner` — Invited team members and partners (Professional and Enterprise).
For tools that require the workspace ID, and no workspace ID is provided, ask the user to provide the workspace ID. If the user does not provide the workspace ID, call this first with the createdBy parameter to use the first workspace.
Results are paginated. Use the `cursor` parameter to retrieve additional pages.
Examples:
“List my workspaces” → GET `/me`, then GET `/workspaces?createdBy={me.user.id}&limit=100`
“List my personal workspaces” → GET `/me`, then GET `/workspaces?type=personal&createdBy={me.user.id}&limit=100`
“List all public workspaces” → GET `/workspaces?type=public&limit=100`
Inputs: type, limit, cursor, include, +3 more.
Publishes a mock server. Publishing a mock server sets its Access Control configuration setting to public.
Inputs: mockId (required).
Replaces the contents of a collection using the Postman Collection v2.1.0 schema format. Include the collection's ID values in the request body. If you do not, the endpoint removes the existing items and creates new items.
To perform an update asynchronously, use the `Prefer` header with the `respond-async` value. When performing an async update, this endpoint returns a HTTP `202 Accepted` response.
For a complete list of properties and information, see the Postman Collection Format documentation.
For protocol profile behavior, refer to Postman's Protocol Profile Behavior documentation.
Note:
The maximum collection size this endpoint accepts cannot exceed 100 MB.
Use the GET `/collection-updates-tasks/{taskId}` endpoint to get the collection's update status when performing an asynchronous update.
If you don't include the collection items' ID values from the request body, the endpoint removes the existing items and recreates the items with new ID values.
To copy another collection's contents to the given collection, remove all ID values before you pass it in this endpoint. If you do not, this endpoint returns an error. These values include the `id`, `uid`, and `postman_id` values.
Inputs: collectionId (required), Prefer, collection.
Replaces all the contents of an environment with the given information.
Note:
The request body size cannot exceed the maximum allowed size of 30MB.
If you receive an HTTP `411 Length Required` error response, manually pass the `Content-Length` header and its value in the request header.
Only shared variable values can be modified through the Postman API. A shared variable is an environment variable with its value synced and stored in the Postman cloud, and can be accessed by your teammates in the environment's workspace.
Inputs: environmentId (required), environment.
Runs a Postman collection by ID with detailed test results and execution statistics. Supports optional environment for variable substitution. Note: Advanced parameters like custom delays and other runtime options are not yet available.
Inputs: collectionId (required), stopOnError, abortOnError, environmentId, +5 more.
Search for Postman entities (requests, collections, workspaces, specs, flows, environments, mocks, and documents).
Ownership:
organization — Search within all resources owned by your organization (default).
external — Search within the public Postman network (third-party and community APIs).
all — Search across all scopes.
When to use each ownership value and filters:
Goal | Recommended approach |
Find an internal API (e.g. "our notification service") |
|
Find a trusted API published to the Private Network |
|
Find an internal API in all resources of organization and are visible to the organization only |
|
Find an API by your organization that is made publicly visible |
|
Find a third party publicly visible API (e.g. "Stripe API", "Twilio API") |
|
User says "our APIs", "internal", "team" |
|
Search across all scopes |
|
Element Types:
requests: Search for individual API requests.
collections: Search for API collections.
workspaces: Search for Postman workspaces.
specs: Search for API specifications.
flows: Search for Postman Flows.
environments: Search for Postman Environments.
mocks: Search for Postman Mock Servers.
documents: Search for Postman workspace documents.
Filters:
Use the filters parameter to narrow results. The top-level key must be $and with an array of condition objects. Each condition object must contain exactly one field key.
Supported filter fields:
Field | Operators | Notes |
|
| All element types. |
|
| Requests and collections only. |
|
| Values: |
|
| Boolean. All element types. |
|
| Boolean. All element types. |
|
| HTTP methods (GET, POST, etc.). Requests only. |
|
| Workspaces and collections only. |
|
| Requests only. |
|
| Specs only. |
|
| Flows only. |
|
| Documents only. |
|
| All element types. |
|
| All element types. |
|
| All element types. |
|
| Boolean. Workspaces, collections, requests, specs, flows, environments, mocks, documents. |
|
| Requests only. |
Filter examples:
Private API Network only: {"$and":[{"privateNetwork":{"$eq":true}}]}
Single workspace: {"$and":[{"workspaceId":{"$eq":"ws-abc123"}}]}
Multiple workspaces: {"$and":[{"workspaceId":{"$in":["ws-1","ws-2"]}}]}
Public visibility: {"$and":[{"visibility":{"$eq":"public"}}]}
GET requests only: {"$and":[{"method":{"$eq":"GET"}}]}
Combine conditions: {"$and":[{"visibility":{"$eq":"public"}},{"workspaceId":{"$eq":"ws-abc123"}}]}
Environments in a workspace: {"$and":[{"workspaceId":{"$eq":"ws-abc123"}}]}
Inputs: q, limit, cursor, filters, +2 more.
Syncs a collection generated from an API specification. This is an asynchronous endpoint that returns an HTTP `202 Accepted` response.
Note:
This endpoint only supports the OpenAPI 2.0, 3.0, and 3.1 specification types.
You can only sync collections generated from the given spec ID.
Inputs: specId (required), collectionUid (required).
Syncs an API specification linked to a collection. This is an asynchronous endpoint that returns an HTTP `202 Accepted` response.
Note:
This endpoint only supports the OpenAPI 2.0, 3.0, and 3.1 specification types.
You can only sync collections generated from the given specification ID.
Inputs: specId (required), collectionUid (required).
Updates a request in a collection. For a complete list of properties, refer to the Request entry in the Postman Collection Format documentation.
Note:
You must pass a collection ID (`12ece9e1-2abf-4edc-8e34-de66e74114d2`), not a collection(`12345678-12ece9e1-2abf-4edc-8e34-de66e74114d2`), in this endpoint.
This endpoint does not support changing the folder of a request.
This endpoint acts like a PATCH method. It only updates the values that you pass in the request body.
Events (`events`). Events are merged by `listen` type, so you only need to send the event you are changing. A `listen` type you omit keeps its stored script. To empty a script, send its event with `script.exec` set to `[]` — sending `events: []` changes nothing.
Inputs: requestId (required), collectionId (required), url, auth, +11 more.
Updates a mock server.
Resource: Mock server entity associated with a collection UID.
Use this to change name, environment, privacy, or default server response.
To activate a server response, set `config.serverResponseId` to the server response's `id`. Pass `null` to deactivate.
Inputs: mockId (required), mock.
Updates a file for an OpenAPI or protobuf 2 or 3 specification.
Note:
This endpoint does not accept an empty request body. You must pass one of the accepted values.
This endpoint does not accept multiple request body properties in a single call. For example, you cannot pass both the `content` and `type` property at the same time.
Multi-file specifications can only have one root file.
When updating a file type to `ROOT`, the previous root file is updated to the `DEFAULT` file type.
Files cannot exceed a maximum of 10 MB in size.
Inputs: specId (required), filePath (required), name, type, +1 more.
Updates an API specification's properties, such as its name.
Inputs: name (required), specId (required).
Updates a workspace's property, such as its name or visibility.
Note:
This endpoint does not support the following visibility changes:
`private` to `public`, `public` to `private`, and `private` to `personal` for Free and Solo plans.
`public` to `personal` for team users only.
There are rate limits when publishing public workspaces.
Public team workspace names must be unique.
Inputs: workspaceId (required), workspace.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 42 tools
Most tools are resource/action specific, but the spec and collection lifecycle has overlapping operations such as generateSpecFromCollection, generateCollection, getGeneratedCollectionSpecs, getSpecCollections, syncCollectionWithSpec, and syncSpecWithCollection. Descriptions help, but an agent can easily misselect among these closely related tools.
Names are consistently camelCase and mostly follow predictable verb-noun patterns such as createCollection, getWorkspace, updateMock, and runCollection. Minor deviations exist, including putCollection/putEnvironment instead of update*, and getAllSpecs versus other get* variants.
The 42-tool surface is large for an agent-facing MCP server and includes many fine-grained operations across collections, specs, mocks, environments, and workspaces. While Postman is a broad product, this count feels heavy and increases selection burden.
The surface covers many create/get/update workflows across major Postman entities, but delete operations are conspicuously absent for collections, requests, responses, environments, mocks, specs, and workspaces. Agents can perform core creation and reading tasks but will hit dead ends for cleanup and full lifecycle management.
Connect any AI agent to 1,000+ apps and 27,000+ actions through one remote MCP server (OAuth).
MCP server connecting AI agents to 100+ apps (Gmail, Slack, Notion, GitHub) via one-click OAuth.
Build and run AI workflows, agents, data tables and 1000+ API integrations from any MCP client.
Discover and call 10,000+ production APIs from one MCP server. Pay-per-call billing for AI agents.
