Smiddy v0.2.0 - Token Usage Baseline

Smiddy v0.2.0 - Token Usage Baseline

This release introduces a Smiddy-native Token Usage Baseline, allowing users to estimate the token footprint of Smiddy-controlled artifacts before making token-efficiency or compression changes.

Added

• Added .smiddy/tools/tokens/token_report.py, an optional Python utility for estimating token usage.
• Added Markdown and JSON token usage reports under .smiddy/tools/tokens/reports/.
• Added category-based token breakdowns for context, governance, prompts, specs, docs, and other artifacts.
• Added largest-file analysis to identify the biggest contributors to context size.
• Added .smiddy/prompts/setup/token-usage.md to guide AI assistants through baseline analysis.

Documentation

• Updated README with Token Usage Baseline instructions.
• Added clarification that token usage is estimated, not exact provider billing.

Notes

This release does not implement token compression yet. It provides the measurement baseline needed to decide whether compression or context optimization is worth implementing.

Smiddy v0.1.0 — First Public Release

Smiddy v0.1.0 — First Public Release

Smiddy is a tool-agnostic SDLC autonomous pipeline framework for running structured, phase-driven software delivery with AI coding assistants.

This first public release introduces the foundation for controlled AI-assisted development: structured specs, persona-driven phases, governance artifacts, quality gates, and adapter support for Claude Code and GitHub Copilot.

What’s included

• Phase-driven SDLC pipeline:
  • 01 Requirements
  • 02 Design
  • 03 Build
  • 04 Review
  • 05 Docs
• Tool-agnostic prompt architecture.
• Claude Code adapter.
• GitHub Copilot adapter.
• Governance model for:
  • Decisions
  • Standards
  • Gates
• Gate checks before phase execution.
• .pipeline-state.yml tracking.
• Resume support for interrupted workflows.
• Install script for adding Smiddy to existing projects.

Why it matters

Smiddy is not just an AI code-generation helper. It is a structured delivery framework for using AI with control, consistency, and engineering governance.

Known limitations

• Manual prompt-driven execution.
• No standalone package distribution yet.
• No full CLI runtime yet.
• No automated release pipeline yet.

Recommended next version

v0.2.0 should focus on the Validation Rules layer.