Skills · Community project

yichen-skills

A Yichen-maintained Agent Skills collection for web research, known-link archiving, authorized bookmark export, Markdown memory, media transcription, X drafting and slicing, and local WeChat or WeCom workflows.

1.9k StarsPersonal learning and non-commercial usePythonUpdated 1 day ago
01

Project overview

yichen-skills is a broad collection rather than a single-purpose Skill. The referenced X post highlighted eight capability groups; by this review, the repository README listed 22 entries spanning public-web research, known-link archiving, read-only private bookmark export, Markdown/Obsidian memory, ASR, X Article drafting, X content slicing, and local WeChat or WeCom workflows. The repository explicitly separates search from download authorization and requires exact, current-task approval for sensitive data and signed-in sessions.

02

Core capabilities

01

Layered web research

yichen-web-research orchestrates multi-stage work while child Skills handle public discovery, known-link archiving, private bookmark export, and ASR. The repository requires original-source review and explicit coverage limitations.

02

Creator content pipeline

Components can summarize conversations, upload Markdown into X Article drafts, create verified X slices and videos, transcribe media, and hand rough cuts to Jianying or CapCut for finishing.

03

Markdown-first memory

yichen-agent-memory uses Markdown or Obsidian as the source of truth, SQLite/FTS as a fast index, and optional semantic retrieval with reconcile, closeout, and audit workflows.

04

WeChat and WeCom workflows

The collection includes local chat and bookmark export, Official Account archiving, read-only WeCom snapshots, and official-CLI document, todo, meeting, and calendar operations. Their system and authorization requirements differ.

05

Safety and authorization routing

Anonymous public routes are preferred. Browser sessions, private bookmarks, credential capture, certificate trust, proxy changes, and app modification require explicit current-task confirmation.

03

Installation and usage

Start with one component that directly matches your task and install only its runtime requirements. At the repository level, clone the project and copy selected component folders unchanged into your host's Skills path. Use an exact npx or Codex plugin command when that component documents one.

AI AGENT INSTALL

Let an AI Agent install it

Send this prompt to Codex, Claude Code, or another AI agent that can work with your local environment.

Help me evaluate and install the single yichen-skills component that best fits my current task. Repository: https://github.com/mcncarl/yichen-skills. First read the root README, LICENSE, THIRD_PARTY_NOTICES.md, and the target component's SKILL.md or README. Explain the personal non-commercial license, supported host, dependencies, data access, and minimum permissions. Do not install the entire collection and do not automatically read signed-in browser state, chat databases, private bookmarks, or keys. Confirm that the target directory is absent or ask before overwriting it, install only necessary dependencies, and run the component's minimal read-only verification. Ask first if credentials, additional permissions, certificate trust, proxy changes, app re-signing, paid APIs, or file replacement are required.
01Before you start
  • Codex CLI, Claude Code, or another host that loads local SKILL.md files
  • Git plus Python 3.9+ or Node.js 18+ when required by the selected component
  • Only the selected component's Playwright, ffmpeg, external CLI, signed-in account, or API credentials
  • Explicit approval before private reads, app changes, certificate trust, proxy changes, or paid API calls
02Copy the install command or configuration
Project URL: https://github.com/mcncarl/yichen-skills
Copy the URL and follow the steps below to complete setup.
03Complete the setup steps
  1. 1
    Choose the smallest component

    Pick one entry from Included Skills that directly serves the current task. Record its path, host, dependencies, data scope, and approval gates instead of treating the collection as one trusted package.

  2. 2
    Review licensing and risk

    Read the root license and component documentation. Obtain written permission for commercial or internal business use, and current-task approval for sessions, private data, certificates, proxies, paid services, or app changes.

  3. 3
    Install from the documented entry point

    Clone the official repository and preserve the selected folder name in the host Skills directory. When the component documents an npx or Codex plugin command, use it exactly and ask before replacing an existing directory.

  4. 4
    Add only required dependencies

    Install only the Python, Node.js, Playwright, ffmpeg, or third-party CLI dependencies for the selected component. Keep credentials in environment variables or secure host storage and out of Git and logs.

  5. 5
    Run a read-only verification

    Check Skill discovery and then run the component's validator or smallest read-only task. Report structure, dependency availability, and actual functional behavior separately.

How to verify the setup

Confirm that the selected SKILL.md is discovered by the host, then run its documented minimal read-only check. For example, the research family can run `python3 yichen-web-research/scripts/validate_family.py` from the repository root. A structural or dependency check does not authorize a private read or paid request.

Before using it
  • Commercial, client, course, or internal business use requires explicit written authorization
  • Authorization for one signed-in session or private dataset must not be reused for another platform or task
  • yichen-social-bookmarks-exporter is a compatibility alias; the maintained implementation is yichen-bookmarks-export
  • The repository's only image is a donation QR code, so this page intentionally has no body media
04

Use cases

SCENARIO 01

Evidence-backed web research

Discover candidates, open original sources, document coverage, and archive or transcribe only when the user supplies an exact scope.

SCENARIO 02

Creator asset processing

Turn confirmed articles, media, or Markdown into archives, transcripts, X drafts, image slices, or video deliverables.

SCENARIO 03

Local long-term memory

Keep editable long-term memory in Markdown or Obsidian and use local indexing for recall and reconciliation.

SCENARIO 04

Personal WeChat asset organization

With data-owner authorization and local safeguards, organize chats, Moments, Favorites, Official Account articles, or WeCom snapshots in read-only workflows.

05

Assessment

Based on the public README, component documentation, and license, the collection's strongest design choice is separating discovery, download, private access, and paid calls into different authorization layers. Many workflows also document verification and no-overwrite behavior. The tradeoff is breadth: platform logins, system dependencies, privacy risks, and a restrictive license must be reviewed per component. This assessment is based on public materials only; the Skills were not installed or run.

Why it may be useful

  • Research, archiving, bookmarks, and ASR have distinct responsibilities and authorization gates
  • Codex and Claude Code are documented, with exact installation or validation commands for selected components
  • The README explicitly keeps credentials, cookies, chat records, and database keys out of Git
  • The repository is recently maintained and preserves upstream attribution and license notices

What to know first

  • The custom personal non-commercial license limits team, company, client, and commercial use
  • The broad component set requires individual auditing rather than bulk trust
  • Some routes involve database decryption, signed-in sessions, proxies, certificates, or app re-signing and therefore carry higher risk than text-only Skills
  • This review verified metadata only and did not test installation or functionality
06

README

yichen-skills


Overview

A Yichen-maintained Agent Skills collection for web research, known-link archiving, authorized bookmark export, Markdown memory, media transcription, X drafting and slicing, and local WeChat or WeCom workflows. A skill collection for creators who want to streamline writing, X content slicing, X Articles draft publishing, WeChat digital-asset capture, and local workflows with Claude Code and Codex.

Getting started

  • Start with one component that directly matches your task and install only its runtime requirements. At the repository level, clone the project and copy selected component folders unchanged into your host's Skills path. Use an exact npx or Codex plugin command when that component documents one.
  • The README currently lists 22 workflow entries, including independent Skills, routers, a compatibility alias, and one Codex plugin.
  • The research family separates discovery, known-link archiving, private bookmark export, and ASR. Search results do not automatically authorize downloads.
  • Confirm that the selected SKILL.md is discovered by the host, then run its documented minimal read-only check. For example, the research family can run `python3 yichen-web-research/scripts/validate_family.py` from the repository root. A structural or dependency check does not authorize a private read or paid request.

Configuration

Project URL: https://github.com/mcncarl/yichen-skills
Copy the URL and follow the steps below to complete setup.
Read the complete README on GitHub