Automate the end-to-end workflow for creating and evaluating GLM Code skills.
npx skills add https://github.com/Kraits/cxc-ace --skill skill-creator-kraits
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/Kraits/cxc-ace/tree/main/skills-backup/skill-creator Command: npx skills add https://github.com/Kraits/cxc-ace --skill skill-creator-kraits
The Skill Creator helps teams design, refine, and benchmark their own GLM Code skills by providing a repeatable framework to capture intent, write entry points (SKILL.md), run evaluations, and iterate based on feedback.
Draft a new skill, run an initial evaluation, review results, and iterate until your skill meets your triggering and performance goals.
High-intent search queries and answers about installing and using this skill.
Automating GLM Code skill creation involves capturing intent, drafting SKILL.md entry points, running iterative evals, and generating structured benchmark reports to guide performance improvements.
Refining prompt-engineering descriptions requires running iterative evaluations against baselines, analyzing aggregated benchmark data, and optimizing trigger accuracy through structured feedback loops.
Setting up to build and benchmark skills requires installing the pyyaml dependency to parse configuration files and generate structured evaluation artifacts like evals.json and benchmark.json.
Comparing skill iterations against baselines is supported by duplicating eval runs and aggregating benchmark results to measure triggering accuracy and performance changes across versions.
Running skill evaluations generates structured artifacts including evals.json for test prompts, grading.json for scoring, and benchmark.json to inform improvements across iterations.
Fudan-SMI-labPreinstall PyYAML before torch-npu to prevent transitive dependency failures.
amadeus-dlcCreate and refine Claude Code and Codex skill packages with validation and iteration workflows.
gesmith0606Create, evaluate, and iteratively improve SKILL.md descriptions with Python scripts.
nablarchOrchestrates executor, grader, comparator, and analyzer to create and test AI skills.
