Skip to content
Back to skills

Infocard X Content Tracing

ASecurity

X content to upstream project/author tracing.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 19, 2026
toolsgogitsecurity

Works with

  • cli

Security analysis

A100/100

Scanned October 19, 2026

npx -y skills add ccwq/infocard-pub --skill infocard-x-content-tracing --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Infocard X Content Tracing?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Infocard X Content Tracing
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/ccwq-infocard-x-content-tracing/badge)](https://www.skillsdirectory.com/skills/ccwq-infocard-x-content-tracing)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: infocard-x-content-tracing
description: "X content to upstream project/author tracing."
version: 1.0.0
author: Hermes Agent
license: MIT
---

# X Content → Infocard Author Tracing

## Trigger

X/Twitter post contains: a code snippet, Rules block, prompt text, implementation guideline, or project recommendation — and the X post is a **mediator** (the author shared it), not the canonical source.

## Goal

Trace the actual upstream project/author, then author the card as a **project card**. The X post provides the discovery signal; the card credits the project/author as primary source.

## Step 1 — Identify canonical upstream

| Content type | Canonical source | Example |
|---|---|---|
| Rules/Prompts file | GitHub repo README/raw | `global.md`, `AGENTS.md`, implementation principles |
| Tool/framework | GitHub repo | DeepSeek Harness plugin, coding agent tool |
| Project collection | GitHub repo | awesome-list, curated toolkit |

**Search for Rules/Prompts**:
```
site:github.com "<unique phrase from the content>"
# e.g. "Fix the root cause with the smallest clear and debuggable change"
```

**Search for tools/projects**:
```
<project-name> GitHub stars 2026
```

## Step 2 — Extract author identity

From GitHub profile:
- Handle + display name
- Join date, follower/repo counts
- Related projects (often a trilogy)
- Stated motivation (README "项目初衷")
- License

**Verified example — `xstongxue/best-rules`**:
- Handle: `xstongxue` / 小帅同学 · Joined: 2024-05 · Followers: 27, repos: 16
- Related: best-prompts, best-skills
- Motivation: "在 Vibe Coding 时代,好的 Prompt 决定 AI 的输出质量"
- License: MIT

## Step 3 — Card structure for Rules/Prompts content

Card is about the **project**, not the X post:
1. Hero: name, stars, license, tagline
2. **Author block** (prominent, near top): GitHub handle, profile stats, related projects
3. Original prompt/Rules text (verbatim code block)
4. Project overview
5. With/without comparison (concrete examples)
6. How to use
7. Change log section if updating existing card

## Step 4 — Anti-patterns

- **Do NOT** publish a Rules card as anonymous content — author is first-class value
- **Do NOT** treat the X post as the source — it is discovery, not canonical
- **Do NOT** skip GitHub search for recognizable Rules/Prompts text — upstream repo almost always exists
- **Do NOT** guess the author — always verify from GitHub profile

## Step 5 — DSh/Plugin ecosystem cards

Include: plugin count, license (often CC0), official CLI install command, security caveat.

**Example — `awesome-dsh-plugin`**:
- Stars: 7,841 · License: CC0 · CLI: `dsh plugin add <owner/repo>`
- Security: plugins run locally, not sandboxed by Tool Approvals

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.