Community写作与编辑github.com

rogerchappel/skill-trace-sampler

Sample agent run transcripts into safe skill fixture packs

skill-trace-sampler 是什么?

skill-trace-sampler is a Claude Code agent skill that sample agent run transcripts into safe skill fixture packs.

兼容平台✓Claude Code~Codex CLI~Cursor
npx skills add rogerchappel/skill-trace-sampler

Installed? Explore more 写作与编辑 skills: steipete/notion, langchain-ai/langchain, bytedance/podcast-generation · View all 6 →

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Skill Trace Sampler

Use this skill when an agent builder needs safe, representative examples from a completed agent run for SKILL.md examples, regression fixtures, demos, or PR evidence.

Inputs

  • Plain text transcript, JSONL transcript, or markdown run notes.
  • Optional --max-per-category to keep the pack small.

Side-Effect Boundaries

  • Reads only the files passed on the command line.
  • Writes only when --out <path> is provided.
  • Does not call external services, execute transcript commands, or inspect live accounts.

Approval Requirements

No approval is required for local parsing. Approval is required before sharing sample output outside the workspace, because transcripts may include sensitive context even after redaction.

Validation

Run npm test, npm run smoke, and inspect the generated redaction notes before using samples in public artifacts.

相关技能