Skip to main content

structured-artifact-inspection

Use for unknown-size, large, generated, or high-output structured artifacts; avoid for small known config files or jq-style queries.

来源信息

仓库
Rycen7822/Structured-Artifact-Viewer
最近来源活动
2026年5月9日 05:26
检测到的 SKILL.md 语言
英语
星标
1
分支
0

安装方式

默认使用会先检查来源的 Prompt;你也可以切换为直接命令,或下载本地副本。

检查来源文件

决定是否安装前,请先阅读 SKILL.md,以及 SkillsMP 当前展示的配套文件。

正在显示 SKILL.md

SKILL.md
来源说明 · 只读预览
name
structured-artifact-inspection
description
Use for unknown-size, large, generated, or high-output structured artifacts; avoid for small known config files or jq-style queries.
# Structured Artifact Inspection Use this skill only for bounded inspection of structured artifacts when raw output could waste context: unknown-size or large JSON, JSONL/NDJSON, CSV/TSV, Parquet metadata, and generated tool artifact files. Do not use this for search, transformation, calculation, validation, editing, or precise jq-style querying. This plugin is a compact viewer, not a jq replacement. For small known files, especially ordinary config files, read them directly unless they contain long records or large fields. Prefer the MCP tool `structured_artifact_viewer` when available. Use the CLI `codex-view` only as a fallback or when the user explicitly asks for shell commands. Workflow: 1. Inspect the file shape with `op: "sniff"` or `codex-view sniff FILE`. 2. Summarize structure with `op: "summary"` or `codex-view summary FILE`. 3. Preview selected fields with `op: "select"` or `codex-view select FILE --fields name,status,score --limit 10`. 4. Never print full JSONL records by default; one record may be tens of kilobytes. 5. If exact full content is required, first prove the file is small or write raw output to a file and report only the path plus a short summary. Avoid raw `cat`, `head`, `tail`, `sed`, `nl`, and `python -m json.tool` on these files unless the user explicitly asks for raw content.
在 GitHub 查看