Skip to content

CHANGELOG.md

Latest commit

 

History

History
40 lines (33 loc) · 2.29 KB

CHANGELOG.md

File metadata and controls

40 lines (33 loc) · 2.29 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

1.2.0 - 2026-07-15

  • Clarified proportional rigor as a cross-cutting constraint rather than a thirteenth principle or a mandatory tier system.
  • Added the smallest-decision-changing evidence rule and an explicit stop condition for additional validation.
  • Clarified that user-reported symptoms are high-value evidence while proposed causes remain hypotheses to verify independently.
  • Added source-backed evidence from invocation economics, fail-fast and early-exit workflows, hard/soft dependency policy, YAGNI architecture scoping, pruning, and fact/decision boundaries.
  • Synchronized the English source documents, Chinese mirrors, installable skill, evidence boundary, case study, repository-engineering guide, notice, and plugin metadata.

1.1.0 - 2026-07-15

  • Added the source README's “Skills for Real Engineers” framing and its four engineering failure themes to the English and Chinese READMEs.
  • Added a source-backed repository-engineering study covering lifecycle, manifests, harness metadata, configuration adapters, human documentation, Changesets, and release automation.
  • Added a dedicated deprecated/in-progress evolution study with recorded-versus-interpreted labels.
  • Added a separate Skill Creator Pro case study covering the OpenAI baseline, independent extensions, three real-skill studies, release evidence, and limitations.
  • Extended the methodology, evidence map, examples, single-file guidelines, and installable skill with a conditional repository-engineering branch.
  • Documented that the Matt source snapshot contains no Python compilation gate, repository-wide validator, universal lint/test suite, or numeric SKILL.md limit.
  • Updated Claude plugin and marketplace metadata to version 1.1.0.
  • Added a dependency-free repository validator, four regression tests, and Windows/Linux validation CI for the study repository's own release contract.

1.0.0 - 2026-07-14

  • Published the twelve Matt Pocock-inspired skill-writing principles.
  • Added English and Simplified Chinese README, methodology, evidence map, and examples.
  • Added a single-file guideline and an installable Agent Skill.
  • Pinned every source claim to mattpocock/skills@66898f6 and documented interpretation boundaries.