About this skill

Problem

When integrating multiple wm-* skills and role-* scenarios into an Agent, teams often run into version drift, repeated auth trial-and-error, treating Roles as executable skills, and leaking credentials into skill packages. SkillHub 管家 centralizes operations: it discovers the remote catalog, installs and upgrades official packs, validates frontmatter and sidecars, and routes requests to the correct Role or sub-skill.

How It Works

Unified entrypoint: use wm.sh discover and wm.sh run instead of hand-writing curl, OAuth flows, or feedback URLs.
Install and upgrade: prefer official pack/catalog paths, keep credentials, token cache, and workspace mirrors under ~/.winmale/, and preserve user-level files while backing up old packages.
Routing and acceptance: match requests to the four scenario Roles (ticker check, portfolio tracking, company deep read, financial analysis) and L1 skills; validate SKILL.md, _skillhub_meta.json, and agents/openai.yaml after install/update.
Errors and feedback: return human-readable failure causes and landing pages, file POST /v1/feedback or feedback.create when needed, and avoid blind retries or claiming issues are fixed.

Boundaries

Best for engineers maintaining official Winmale SkillHub skill and role assets. It is not a substitute for third-party data access, and machines with existing credentials should not be reinstalled with first-run key prompts.

Screenshots

Use Cases

When first wiring Winmale skills into Cursor or Claude Code, install the official pack and validate frontmatter, sidecars, and display names.
When local wm-* skills need upgrades, compare catalog semver versions, replace official packs, and preserve credentials, cache, and workspace files.
When users request a company research memo, route the request to the company deep-read Role instead of only calling a company card or quote snapshot.
When hitting SCOPE_DENIED, exhausted quota, or invalid keys, follow error recovery to surface the cause, landing page, and feedback channel.

Best For

Agent engineers maintaining Winmale skills in Cursor, Claude Code, or WorkBuddy who need unified install, upgrade, and pack validation.
Engineers orchestrating research agents who need to route ticker checks, company deep reads, watchlists, and screening requests to the right Role.
Technical owners operating financial data platforms who need credential isolation, token cache handling, error recovery, and feedback reporting.
Developers using SkillHub across clients who need to adapt skill roots, Chinese display names, and Codex or WorkBuddy sidecars.

Related Skills

Seedance 2.0 Video Prompt Writing Guide

@user_d5c3472c/seedance-prompt

A guide for writing Seedance 2.0 video prompts, covering @ references, camera moves, time segments, audio direction, and common pitfalls.

AI Agent 2026.08.30 ↓ 880

AI Daily Self-Review System

@user_fe828a0e/self-reivew-pro

Collects daily conversations, task logs, and errors, then runs a nine-module review to surface blind spots, generate actions, and update memory.

AI Agent 2026.08.30 ↓ 1.6K

Permanent Memory Auto-Archiving System

@user_6bedba2d/ybdsjy

Local long-term memory for AI agents that appends key facts and daily logs, supports cross-session context and search recall, and auto-archives old diaries to gzip files.

AI Agent 2026.08.30 ↓ 1.2K

Auto-Evolving Agent

@user_dfbdc5ab/auto-evolving-agent

A hybrid-evolution agent for OpenClaw developers that offers tiered-confirmation and rollback-safe capability optimization.

AI Agent 2026.08.30 ↓ 999

Winmale SkillHub Manager