Socket Help

Connect your agent

170 tools in 12 categories: HTTP and webhooks, email, mocks and OAuth, files, monitors, memory, authenticators, jobs and status pages, test data and media, developer utilities, DNS and network. Start with search_tools to discover them.

Open the setup guide →

claude mcp add --transport http socket-help https://socket.help/mcp

Or call the HTTP API

Format JSON without an account. The response contains the tool ID and a text result. Invalid input returns a structured error with a request ID.

Read the quickstart →

curl https://socket.help/api/tools/json/run \
  -H 'Content-Type: application/json' \
  -d '{"input":"{\"hello\":true}","options":{"action":"format"}}'

Get started

Connect an agent and use email, memory, monitors and approvals Connect a coding agent through browser approval, receive signup codes, save workspace memory, monitor jobs and wait for human answers through MCP, HTTP or SDK.Quickstart: your first socket.help API request Make your first socket.help API call with curl and no account: format JSON, create a webhook inbox, send a test callback, then add a workspace API key.Authentication, API keys and resource tokens How socket.help authenticates requests: anonymous use, scoped workspace API keys, browser sessions, and the separate inspection, management and access tokens.Limits, rate limits and error responses socket.help plan allowances, rate limits, request sizes and retention, plus the JSON error format, status codes and how to handle Retry-After on 429 and 503.

Resources

Test OAuth and OpenID Connect sign in Create an OAuth and OpenID Connect test issuer with synthetic users, S256 PKCE, signed ID tokens and controlled failures. Inspect each request in your workspace.Test a WebSocket client with a hosted mock Create a temporary WebSocket mock with echo, scripted replies and close codes. Inspect its traffic and test public WebSocket connections from any interface.HTTP echo endpoints Create a private temporary HTTP echo endpoint and inspect request metadata and bodies through curl, SDK, CLI, SSH and MCP with no retained request history.Stateful CRUD mock API Create a seeded collection, call its record endpoints, manage revisions and sharing, and reset live data through HTTP, SDK, CLI, SSH or MCP.Webhook inboxes: capture and inspect HTTP callbacks Create a temporary webhook URL with the socket.help API, receive callbacks from any service, and read captured methods, headers, query values and bodies.Mock REST APIs: configured routes with static or templated responses Create a temporary mock REST API on socket.help with up to 20 routes, path parameters, response templates, conditions, delays, rate limits and streaming.Temporary file sharing, private links and one-time downloads Upload a file up to 10 MiB with the socket.help API, share it privately or publicly, limit downloads with maxDownloads, and create one-time download links.Mock MCP servers for testing agents and MCP clients Create a temporary mock MCP server on socket.help with up to 10 tools, input schemas, static responses, errors and delays, served over Streamable HTTP.Sharing and access: public and private resources How socket.help visibility works: private resources require an X-Access-Token, public ones accept anyone with the URL, and access tokens can be rotated.Resource lifetimes, renewal and persistent leases How long socket.help resources last: the one-hour default, ttlSeconds, renewal up to your plan’s maximum retention, persistent paid leases and cleanup.

Interfaces

API test scenarios: format, execution and reports Run reusable API scenarios through HTTP, SDK, CLI, SSH or MCP. Configure requests, variables, checks and captures with separate secrets and bounded execution.Run browser utilities through the API Run socket.help utilities over HTTP with POST /api/tools/{id}/run: find tool IDs with GET /api/tools, pass input, options and files, and read the output.socket-help command line interface Use the socket-help CLI to run utilities from a pipe, create webhook inboxes, upload and download files, and call any /api/ endpoint with JSON output.TypeScript SDK for socket.help Use the @socket-help/sdk TypeScript client to run utilities, create inboxes, mocks, files and mock MCP servers, change sharing, renew and handle typed errors.SSH gateway for socket-help commands Run socket-help CLI commands over SSH with your API key as the password. No shell, forwarding, SCP or SFTP: one command per session with JSON output.MCP endpoint for AI agents Connect an AI agent to the socket.help MCP endpoint at /mcp to run utilities and manage webhook inboxes, mock APIs, files and mock MCP servers over HTTP.

Reference

HTTP API reference Every endpoint with parameters, request bodies, responses, and runnable examples.OpenAPI JSON The machine readable specification, for code generators, clients, and agents.

Frequently asked questions

Do I need an account to use the socket.help API?

No. Browser utilities and anonymous temporary resources work without an account. An account adds a workspace, longer retention within plan limits, and scoped API keys.

Where is the OpenAPI specification?

The full OpenAPI document is served at /api/openapi.json, and the interactive reference is at /docs/reference/.

Can AI agents use socket.help?

Yes. Agents can call the HTTP API like any script, or connect to the socket.help MCP endpoint to discover and run tools and manage resources.

Is there a CLI or SDK?

Yes. A TypeScript SDK and a command line tool wrap the same HTTP API, permissions, and allowances. See the Interfaces pages for setup.