Pyvider absorbs the gRPC transport, mTLS handshake, and Plugin Protocol v6 serialization — so you write resource logic, not wire protocol.
Python 3.11+ with attrs, type hints, and async/await. Use the full Python ecosystem in your providers.
Register providers, resources, and functions with simple decorators. The hub handles discovery automatically.
Full Terraform Plugin Protocol v6 implementation. Compatible with Terraform and OpenTofu out of the box.
Built on modern async/await patterns for high-performance concurrent resource operations.
Full MyPy compatibility. Schema validation catches configuration errors before Terraform does.
First-class testing support with pytest integration and a built-in test mode for unit testing providers.
Build structured, non-blocking guidance in Python today. See the checked OpenTofu 1.13.0-rc1 experimental validation lane and direct proof for all seven Pyvider hooks.
The scaffolding and protocol are handled. What's left is your resource — schema definition and CRUD logic, in Python.
Resources, data sources, and provider functions — tofu init through tofu apply.
The full surface area of Terraform Plugin Protocol v6. Decorate your classes — the hub wires them to the plugin RPC layer automatically.
Configuration, authentication, and initialization for your provider.
CRUD-managed infrastructure. Full create, read, update, delete lifecycle.
Read-only queries. Pull existing infrastructure state into your Terraform configs.
Custom logic callable from HCL. Transform and compute values inside Terraform configurations.
Short-lived write-only resources for secrets, tokens, and temporary credentials.
Flavorpack packages pyvider providers into self-contained executables — no Python required on the end user's system.
Build your Terraform provider using pyvider. Decorate resources, implement CRUD — pure Python.
Run flavor pack to bundle your provider, Python runtime, and all dependencies into a single .psp file.
Terraform executes the binary directly. No Python installation required. pyvider itself supports Linux, macOS, FreeBSD, and more.
The full pyvider conformance suite — every resource, data source, and function exercised end-to-end.
The same suite across every supported Terraform and OpenTofu version.
Core libraries and tools that power the Pyvider ecosystem.
The framework. Decorator-based API for building Terraform providers in Python — protocol, schema, and hub all included.
github.com/provide-io/pyvider →Pure-Python implementation of HashiCorp's go-plugin protocol with full mTLS support and async-first design.
github.com/provide-io/pyvider-rpcplugin →Pure-Python implementation of the go-cty type system with strong validation, serialization, and Terraform interoperability.
github.com/provide-io/pyvider-cty →HCL parsing integrated with Pyvider's Cty type system — parse HashiCorp Configuration Language into typed Python values.
github.com/provide-io/pyvider-hcl →Standard reusable component library for Pyvider. Reference implementations and drop-in building blocks for common provider patterns.
github.com/provide-io/pyvider-components →The official reference Terraform provider built entirely in Python with pyvider — a working proof and learning resource.
github.com/provide-io/terraform-provider-pyvider →Package pyvider providers as single-file, self-contained executables. No Python required on the end user's system.
foundry.provide.io/flavorpack →Cross-language conformance test suite for OpenTofu tooling. Validates protocol compatibility across Python and Go implementations.
tofusoup →Terraform provider for querying OpenTofu/Terraform registries and inspecting state files — cross-stack references without remote backends.
tofusoup →Automatic documentation generation for Terraform providers. Schema extraction, capability-first organization, and registry-ready output.
plating →Part 1 of 5: set up a project, define a provider class, and implement a full CRUD resource.
Providers can now offer opt-in, rule-selectable configuration advice from every configurable component. Pyvider 0.8.1 is the first published 0.8 release; 0.8.0 was tagged but stopped by its publication guard before reaching PyPI.
Stronger schema and returned-state validation, safer lifecycle planning, corrected state-store behavior, and hardened protocol diagnostics move failures to the boundary where provider authors can act on them.