Design, test, and refine GLM-based skills with structured SKILL.md files.
npx skills add https://github.com/rcstrue/php_payroll --skill skill-creator-rcstrue
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/rcstrue/php_payroll/tree/main/php_payroll/skills/skill-creator Command: npx skills add https://github.com/rcstrue/php_payroll --skill skill-creator-rcstrue
The Skill Creator provides a repeatable, documentation-driven workflow to design, test, and refine GLM-based skills, reducing guesswork and increasing reliability.
Draft a new skill concept, write SKILL.md, add an initial eval set, and run the loop to begin iterating.
High-intent search queries and answers about installing and using this skill.
To create and test GLM skills, define trigger conditions, success criteria, and required inputs, then iterate using evals and benchmarks to optimize performance and produce a ready-to-package skill.
The best way to design a repeatable GLM skill authoring process is using a documentation-driven workflow that translates user intent into structured SKILL.md files with consistent frontmatter and rich body content.
Evaluation-driven iteration works by orchestrating user interviews, running test prompts, evaluating triggers, and performing iterative rewrites to compare progress across benchmark dashboards.
Yes, you need PyYAML to validate and package GLM skills, as the guided packaging and distribution process relies on scripts that parse the structured YAML frontmatter of your SKILL.md files.
Yes, you can use this workflow to benchmark skill performance across iterations by generating eval prompts, tracking results, and comparing progress on dashboards to optimize GLM skill reliability.
Manually authoring GLM skills without an evaluation loop increases guesswork and reduces reliability, lacking the structured trigger evaluations and iterative rewrites needed to optimize performance.
etewe333Coordinate end-to-end Claude skill design from intent capture to SKILL.md drafting and eval-driven iteration.
p-iknowCreate, validate, and package AI agent skills with SKILL.md metadata.
easyspace-aiCreate, evaluate, and optimize Claude Skills with templates and benchmark loops.
FredericoMatsuokaCreate, modify, and benchmark Opensquad skills with SKILL.md generation.
