ExplainThisRepo v0.20.1

What's Changed

• fix: support android (termux) in launcher target resolution by @calchiwo in #209

Full Changelog: v0.20.0...v0.20.1

ExplainThisRepo v0.20.0

Major architecture shift.

Python is now the only core. Node is a launcher.

What shipped

• Python owns all runtime logic

• Node CLI reduced to binary launcher

• Native binaries built via PyInstaller

• Multi-platform support: macOS, Linux, Windows

• npm now distributes prebuilt binaries

• pip continues as full Python install

What changed

• Removed Node runtime logic from execution path

• Removed provider logic from Node

• Unified all behavior in Python

• Introduced binary build + staging pipeline

• Added checksum verification in release flow

• Synced versioning across npm and PyPI

How it works now

• CLI runs → Node launcher executes

• Correct binary is selected per platform

• Python core handles everything

• Output is passed through directly

No duplicated logic. No divergence.

Install

npm (no Python required):

npm install -g explainthisrepo

pip (Python source):

pip install explainthisrepo

Source

Discussion: #203
Pull Request: #208

Notes

• Node is infrastructure only

• Python is the product

• Both npm and pip run the same engine

This is not an incremental update.
This is the system reset.

Full Changelog: v0.10.0...v0.20.0

explainthisrepo v0.10.0

What's Changed

• Adds additional CLI entrypoints to improve usability and typing speed. by @calchiwo in #190
• Add CLI aliases for Node.js CLI parity by @calchiwo in #191
• refactor(node-providers): require model in config for ollama and fix validation by @calchiwo in #194
• feat(node-provider): implement anthropic provider by @calchiwo in #195
• feat(node-providers): implement openrouter provider by @calchiwo in #196
• feat(node-provider): Implement groq provider by @calchiwo in #197

Full Changelog: v0.9.6...v0.10.0

explainthisrepo v0.9.6

Add CLI aliases for faster and more flexible usage.

• explain-this-repo for readability
• etr for fast access
• explainthisrepo remains unchanged

All commands map to the same implementation. No breaking changes

Full Changelog: v0.9.5...v0.9.6

ExplainThisRepo v0.9.5

What's Changed

• feat(providers): implement Groq provider by @calchiwo in #186
• chore(deps): bump next from 16.1.6 to 16.1.7 in /apps/web by @dependabot[bot] in #184
• refactor(providers): remove default model and require user-defined model for groq by @calchiwo in #187
• feat: add OpenRouter Provider by @calchiwo in #188

Full Changelog: v0.9.4...v0.9.5

ExplainThisRepo v0.9.4

Add Groq provider and require user-defined model selection.

Removes default model and updates init flow to prompt for model.

Full Changelog: v0.9.3...v0.9.4

ExplainThisRepo v0.9.3

Added Groq Provider

Supports API key + explicit model selection via --llm groq

Full Changelog: v0.9.2...v0.9.3

ExplainThisRepo v0.9.2

Added Anthropic (Claude) Provider

Supports API key + explicit model selection via --llm anthropic

Full Changelog: v0.9.1...v0.9.2

ExplainThisRepo v0.9.1

Add cross-platform standalone binaries.

• macOS (arm64, x64)
• Linux (arm64, x64)
• Windows (x64)

ExplainThisRepo v0.9.0 - Multi-LLM Architecture

ExplainThisRepo is now model-agnostic.

This release introduce a pluggable multi-LLM architecture with support for Gemini, OpenAI, and Ollama.

Users can now select the model backend at runtime using --llm

Highlights

• Pluggable provider architecture

• Runtime model selection via --llm

• Structured multi-provider configuration

• Delegated provider diagnostics

• Clean separation between CLI and vendor SDK logic

Architecture Changes

• Provider Abstraction Layer

• Introduced LLMProvider base contract

• Enforced validate_config, generate, and doctor

• Removed direct Gemini coupling from generation pipeline

New generation flow:

CLI → generate → registry → selected provider

Provider Registry

• Centralized provider resolution

• Config-based default provider selection

• Runtime override via --llm

• Removed global runtime override state

Concrete Providers

Implemented:

• Gemini

• OpenAI

• Ollama (local and cloud-routed)

Each provider now owns:

• Client initialization

• Configuration validation

• Error handling

• Diagnostics

No vendor-specific logic remains in CLI or core modules.

Configuration Changes

• Structured multi-provider config format:

[llm]
provider = "gemini"

[providers.gemini]
api_key = "..."

[llm]
provider = "openai"

[providers.openaii]
api_key = "..."

[llm]
provider = "ollama"

[providers.ollama]
model = "llama3"
host = "http://localhost:11434"

explainthisrepo init now supports selecting and configuring:

• Gemini

• OpenAI

• Ollama

Runtime Model Selection

Models can be selected per command:

explainthisrepo owner/repo --llm gemini
explainthisrepo owner/repo --llm openai
explainthisrepo owner/repo --llm ollama

--llm works with all modes.

Doctor Improvements

• Diagnostics delegated to active provider

• Removed hard-coded Gemini assumptions

• Respects config and runtime overrides

Optional Dependencies

Provider SDKs are installed via extras:

pip install explainthisrepo[gemini]
pip install explainthisrepo[openai]

Ollama requires no additional Python dependency.

⚠ Breaking Changes

Config schema changed from single-provider to structured multi-provider format.

Users must re-run:

explainthisrepo init

after upgrading.

Full Changelog: v0.6.0...v0.9.0