
HAL
Give your coding agent a stdio MCP bridge so it can call REST and HTTP APIs while you build integrations.
Overview
HAL is a MCP server for the Build phase that gives coding agents HTTP API request capabilities over stdio MCP.
What is this MCP server?
- HAL (HTTP API Layer) MCP server exposes HTTP API operations to LLMs over stdio
- Published on npm as hal-mcp (registry v1.0.14) with GitHub source at DeanWard/HAL
- Fits Claude Code, Cursor, and other MCP clients that support stdio transports
- Lets agents fetch, post, and probe endpoints without one-off curl scripts in every session
- Local npm package—no hosted gateway required beyond the APIs you already use
- Server schema version 1.0.14
- npm package identifier hal-mcp with stdio transport
- Single-package MCP server entry in the official server manifest
Community signal: 38 GitHub stars.
What problem does it solve?
Your agent can write code but struggles to reliably execute real HTTP calls against the APIs you are integrating.
Who is it for?
Solo builders who want Claude Code or Cursor to hit REST endpoints, webhooks, or internal services during integration work.
Skip if: Teams that need enterprise API gateways, centralized secrets management, or compliance-grade traffic logging out of the box.
What do I get? / Deliverables
After you register hal-mcp, the agent can invoke HTTP operations as MCP tools while you build and test integrations.
- MCP-registered HTTP API tools available to the agent
- Repeatable agent-driven calls against your REST or HTTP services
- Faster integration debugging without one-off shell HTTP commands
Recommended MCP Servers
Journey fit
HTTP client tooling lands when you are wiring backends, webhooks, and third-party APIs into an agent-driven workflow. HAL is cataloged under integrations because its core job is exposing HTTP request capabilities to the model, not UI or infra provisioning.
How it compares
Stdio MCP HTTP bridge, not a full API gateway or OpenAPI codegen skill.
Common Questions / FAQ
Who is HAL for?
HAL is for indie and solo developers who use MCP-enabled agents and need repeatable HTTP access to backends and third-party APIs while building.
When should I use HAL?
Use HAL during the build and integrations phase when you want the model to call live HTTP endpoints, debug responses, or exercise APIs without manual curl each time.
How do I add HAL to my agent?
Install the npm package hal-mcp, add an MCP server entry with stdio transport in Claude Code or Cursor, then restart the client so the HTTP tools load.