mcpField
L
lunge

lunge

Local
lunge.devDeveloper Tools0Updated 14w ago
RepositoryVisit website

About lunge

lunge is an MCP server that lets AI agents execute and test REST, GraphQL, WebSocket, and SSE requests. It supports one-off calls and reusable YAML or JSON collections, while reducing response size for model context.

Key features

  1. 01REST, GraphQL, WebSocket, and SSE requests
  2. 02Assertions and value extraction
  3. 03Reusable YAML and JSON collections
  4. 04Summarized, context-efficient responses
  5. 05Full payloads retrievable by handle

Use cases

  1. 01Run an API request from an AI agent
  2. 02Test endpoints with assertions
  3. 03Save and rerun request collections

MCP support

It exposes API execution and testing as MCP tools for compatible AI agents. The README describes a local stdio server.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "thuupx-lunge": {
      "command": "npx",
      "args": [
        "-y",
        "lunge-mcp"
      ]
    }
  }
}
Run commandnpx -y lunge-mcp

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npmlunge-mcp1.3.1

Tools

22

Saved tool list · updated 3w ago

ParameterTypeRequiredDescription
envstringNoEnvironment name for variable resolution.
urlstringYesFull URL. May contain {{variables}}.
authobjectNo
auth.inenumNoheader | query
auth.keystringNo
auth.typeenumYesbearer | basic | apikey
auth.tokenstringNo
auth.valuestringNo
auth.passwordstringNo
auth.usernamestringNo
bodyanyNoRequest body (object=json, string=text).
queryobjectNo
assertarrayNoe.g. [{status:200},{jsonpath:'$.x',equals:1},{timeMs:{lt:500}}].
methodstringNoHTTP method (default GET).
extractobjectNoCapture vars: { token: '$.token' }. Use via {{token}} later.
headersobjectNo
bodyTypeenumNoDefault json.json | form | text | raw
timeoutMsintegerNo
verbosityenumNosummary (default, token-efficient), headers, or full. Auto-downgraded to summary when extract is set.summary | headers | full
followRedirectsbooleanNo

Details

From the project README

lunge

An agent-native advanced API client - execute and test REST, GraphQL, WebSocket, SSE, and (later) gRPC requests from any MCP-capable AI agent. No GUI, no manual clicking, no heavyweight desktop app. Think advanced curl, native for agents.

It is a lightweight, token-efficient alternative to Postman/Bruno for the era of AI coding agents. The agent calls tools directly; there are no humans in the loop and no bloated feature surface to pay for.

Why this exists

Pain with Postman/Brunolunge approach
Require human clicking in a GUIAgent-native: everything is an MCP tool call
Postman is heavy, enterprise features are paywalledSmall, open, single local binary + thin MCP layer
Verbose responses blow up the LLM context windowRust core summarizes/compresses output before it reaches the agent
Collections are locked into proprietary formatsPlain YAML/JSON files, git-friendly
Separate tools per protocolOne server for REST/GraphQL/WS/SSE/gRPC

Design pillars

  1. Agent-native - designed for LLM tool calls, not human UIs.
  2. Token-efficient - every response is filtered, truncated, and summarized. Full payloads are spilled to disk and referenced by handle so the context stays small.
  3. Fast & portable - a Rust core (execution + protocols + summarization) exposed to a thin TypeScript MCP layer via napi-rs.
  4. Both ad-hoc and reusable - agents can fire one-off requests or create, save, and re-run declarative collections stored as files.

Documents

DocContents
architecture.mdSystem design, the Rust/TS split, module layout, streaming strategy, FFI boundary
features.mdFull feature list by protocol + cross-cutting features (auth, vars, assertions, token optimizer)
mcp-tools.mdThe MCP tool surface: names, inputs, outputs, examples
collection-format.mdThe declarative YAML/JSON test/collection format
roadmap.mdPhased delivery plan from scaffolding to gRPC and CI

Decisions locked in

  • Stack: Rust core via napi-rs + TypeScript MCP layer (@modelcontextprotocol/sdk).
  • Distribution: local dev tool over stdio transport (used by Cursor/Windsurf/Claude Desktop).
  • Test definition: both ad-hoc agent calls and persisted declarative collections.
  • MVP protocols: REST, GraphQL, WebSocket, SSE. gRPC is a later phase.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.