write-skills and eval-skills

Creates concise operational skills, then evaluates them blindly against golden cases with separate judges.

Parent repository GitHub

Type: skill-authoring skills
Repository: dzhng/skills
Readiness: Usable selectively with caveats; the multi-day autonomous workflow requires capable host tooling, project-specific verification, and human review.
Keywords: 5

Location

Repository path

skills/authoring/write-skills/SKILL.md; skills/authoring/eval-skills/SKILL.md

Invocation

/write-skills; /eval-skills

Setup

Installation / activation

Creating or revising a skill, or testing and hardening one against supplied cases and bars.

Keywords

skill authoringevaluationgolden casesblind runprogressive disclosure

Repository context

dzhng/skills

MIT-licensed collection of 22 README-enumerated Agent Skills for Claude Code and other skill-compatible coding harnesses, centered on an autonomous software-factory loop that explores unknowns, writes and implements living specs, reviews each slice, records agent-made choices, validates visuals, and closes shipped specs into rationale records.

software engineeringrequirements discoverytechnical specificationagent orchestrationcode qualitytestingperformancedocumentationvisual QAWebGPU renderingskill authoringmarketing websites

Open full repository research