Glossary

This glossary defines terms and concepts used throughout the Hatch documentation.

A

Auto-approve: CLI flag (--auto-approve) that automatically approves dependency installations without user prompts, useful for automation scenarios.

Cache TTL: Time-to-live for cached registry data, configurable via --cache-ttl flag. Default is 86400 seconds (24 hours).
Conda: Python package and environment management system used by Hatch for creating isolated Python environments.
Current Environment: The active Hatch environment that commands operate on by default. Set using hatch env use <name>.

Dependency: A package or resource required by another package. Hatch supports multiple dependency types: Hatch packages, Python packages, system packages, and Docker images.
Dependency Installation Orchestrator: Component in hatch/installers/dependency_installation_orchestrator.py that coordinates the installation of multiple dependency types.
Docker Dependency: A Docker image required by a Hatch package, managed by the Docker installer.

Entry Point: The main file that serves as the executable entry point for a Hatch package, typically hatch_mcp_server_entry.py.
Environment: An isolated workspace for managing packages and their dependencies. Hatch environments can optionally include Python environments.
Environment Manager: Core component in hatch/environment_manager.py that handles environment lifecycle and management operations.

FastMCP: A Python framework for building Model Context Protocol servers, used as the base for Hatch package MCP server implementations.
Force Download: CLI flag (--force-download) that forces package download even if the package is already cached locally.

Global Options: CLI options available for all commands, including --envs-dir, --cache-ttl, and --cache-dir.

Hatch: Package manager for Model Context Protocol (MCP) servers that provides environment isolation and dependency management.
Hatch Compliant: Indicates whether a package follows Hatch packaging standards and metadata requirements.
Hatch Dependency: Another Hatch package required as a dependency, managed by the Hatch installer.
Hatch Environment: An isolated environment managed by Hatch for organizing packages and dependencies.
Hatch MCP Server Wrapper: Integration component (hatch_mcp_server) that bridges Hatch packages with MCP server functionality.
Hatch Package: A package that follows Hatch conventions and includes hatch_metadata.json with required metadata fields.
Hatchling: Related project in the Hatch ecosystem that provides additional functionality and tooling.

Installation Context: Object that manages state and context information during package installation processes.
Installer Base: Abstract base class in hatch/installers/installer_base.py that defines the interface for all installer types.

Mamba: Fast, drop-in replacement for conda package manager. Hatch prefers mamba when available for better performance.
MCP: Model Context Protocol - a standard for building AI-powered tools and integrations.
MCP Server: A server implementation that follows the Model Context Protocol standard, typically providing tools and resources for AI applications.
Metadata: Package information stored in hatch_metadata.json that defines package properties, dependencies, and compatibility requirements.

Package: A distributable unit of code that provides MCP server functionality. In Hatch, packages follow specific structure and metadata requirements.
Package Loader: Component in hatch/package_loader.py that loads and validates packages from local directories.
Package Schema Version: Version of the metadata schema used by a package, currently “1.2.0”.
Python Dependency: A Python package installed via pip, managed by the Python installer.
Python Environment: A conda/mamba environment containing a specific Python version and packages, optionally created within Hatch environments.
Python Environment Manager: Component in hatch/python_environment_manager.py that manages Python environments within Hatch environments wrapping around conda/mamba in order to associate python environments with Hatch environments.

Refresh Registry: CLI flag (--refresh-registry) that forces refresh of registry data, bypassing cache.
Registry: Central repository for discovering and downloading Hatch packages.
Registry Explorer: Component in hatch/registry_explorer.py that provides package discovery and search capabilities.
Registry Retriever: Component in hatch/registry_retriever.py that handles package downloads and caching from the registry.

Schema: JSON schema definition that validates package metadata structure. Cracking Shells defines and uses schemas for its package ecosystem. It is hosted on a dedicated repository.
Semantic Versioning: Version numbering scheme using MAJOR.MINOR.PATCH format (e.g., 1.2.0).
System Dependency: A system package installed via OS package managers like apt, managed by the system installer.

Template Generator: Component in hatch/template_generator.py that creates new package templates with standard structure and files.
Tool: A function provided by an MCP server package, defined in the package metadata and exposed through the MCP interface.

Validation: Process of checking package structure and metadata against Hatch requirements and schema definitions.
Version Constraint: Specification of acceptable version ranges for dependencies using operators like >=, ==, <=, !=.

Workspace: Development environment containing multiple related projects and packages, such as the Hatch multi-project workspace.