v0.2.0 — Light Workflow 34-Skill Architecture (extended release line)

v0.2.0 — Light Workflow 34-Skill Architecture (extended release line)

中文发布说明 · Release Receipt

Release line update — 2026-08-30

The v0.2.0 tag was re-pointed to the extended release commit e063753; this release line now carries 34 first-party Skills.

• New 34th package — humanizer (model-invoked): rewrites AI-sounding English or Chinese text so it reads naturally without changing what it says. Adapted from blader/humanizer (2.11.2, MIT) with the pattern book preserved verbatim, plus a thin Chinese adaptation layer (references/zh-adaptation.md: Chinese punctuation overrides, pattern mappings, Chinese AI-vocabulary list, anti-fabrication rule); Chinese vocabulary informed by op7418/Humanizer-zh (MIT). Provenance in the package's ATTRIBUTION.md; full-path admission record in docs/evidence/admissions/humanizer/.
• Post-publication hardening folded into this line: host-agnostic agent-config, user-optional agent-config in implement, and the model-led ask-light advisor refactor with its trusted-host transition policy and regression suite.
• Verification: CI collection-quality PASS on the release commit (run 33292549793); fresh released-repository installation of the re-pointed tag passed — whole collection (34 packages, 257 files byte-identical, CLI 1.5.23) and per-Skill humanizer — see INSTALLATION_VERIFICATION_ADDENDUM.md.

What's new

Light Skills v0.2.0 introduces the complete 33 first-party Skill architecture (expanded from 9 packages in v0.1.6 to 33 packages), covering end-to-end project development workflows, the Socratic clarification engine, execution and review subsystems, and the comprehensive collection advisor ask-light.

Core Architecture Evolution

1. Project Workflow Lifecycle:

   • Standardized lifecycle: project-init → project-clarify → project-spec → project-tickets → implement → project-review → release-workflow.
   • Clear input/output contracts and artifact handoff boundaries at each stage, supporting both end-to-end delivery and single-stage invocation.

2. Socratic Clarification & Decision Engine:

   • Core engine socratic powers independent multi-round questions and option frontier exploration.
   • Powers lightweight clarification clarify, engineering-aware project-clarify, and multi-session decision-map.

3. Execution & Specialist Review Subsystems:

   • agent-config inspects host environment evidence to plan safe execution paths.
   • review-loop provides lightweight review-repair convergence; generic-review and code-review act as read-only specialist reviewers; project-review governs frozen baselines and final verdicts (PASS/FAIL/BLOCKED).

4. Self-Contained Approved Ports & Provenance:

   • 11 approved self-contained Matt Ports with ATTRIBUTION.md and zero upstream runtime dependencies.
   • Incorporated eli5 (from DreambigOu/ELI5) and release-workflow (from LightDevCoder/release-workflow) into unified collection governance with complete migration and retirement records.

5. Collection Advisor:

   • ask-light serves as the read-only router and workflow advisor across all 33 Skills, suggesting next actions based on workspace context.

Installation & Verification

• Interactive Install (Recommended):

  npx skills add LightDevCoder/skills

• Pinned Stable Release:

  npx skills add LightDevCoder/skills#v0.2.0

• Single Skill Example:

  npx skills add LightDevCoder/skills --skill project-review

Verification: Local automated test suites (309 pytest, 27 unittest, compileall, git diff check) and GitHub Actions CI (Run ID 33137041472) all PASS. Fresh isolated environment tests verified generic latest and pinned #v0.2.0 across whole-collection and all 33 individual package install matrices (66/66 PASS).

---

中文说明

完整中文发布说明请访问：RELEASE_NOTES.zh-CN.md