chenwei791129

hermes-plugin-operations

Install, reinstall, migrate, and verify Hermes Agent plugins, especially third-party standalone plugins under ~/.hermes/plugins.

skill-porting-and-adaptation

Audit, port, and adapt external AI-agent skills into Hermes while preserving dependencies, support files, and platform semantics.

Use when operating a Grafana instance from the terminal — listing/searching dashboards, querying datasources, inspecting alert rules, running PromQL/LogQL/TraceQL over metrics/logs/traces, or pulling/pushing dashboards as code (GitOps). Triggers on Grafana, gcx, PromQL, LogQL, dashboard, alerts/告警, 查 Grafana, dashboard 管理, 監控查詢. Use this instead of a Grafana MCP server or raw curl/grafana-backup/grizzly. Self-hosted focused.

tw-trading-agents

台股多 agent 投研分析（TradingAgents 方法論 + FinMind 資料）。當使用者要對某檔台股做完整的多面向投資分析、想要買賣決策建議、或提到「分析台股 2330」「TradingAgents ��股」「台股投研報告」「幫我研究台積電該不該買」時觸發。觸發於股票代碼（如 2330、2317、2454）或公司名（台積電、鴻海、聯發科）且要求深入分析或投資決策的情境。Triggers on "tw-trading-agents", "台股投研", "台股 TradingAgents".

evaluate-skill

Evaluate whether a Claude Code skill improves task outcomes through repeated paired experiments, blind LLM judging, statistical tests, and a self-contained HTML report. Use when the user asks to benchmark, compare, validate, or quantify the effectiveness of a skill. This is a costly explicit-only workflow.

brainstorming

Use before creative or product work: creating features, building components, adding functionality, or changing behavior. Refines intent, requirements, and design before implementation, then hands off to Hermes writing-plans.

finishing-a-development-branch

Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting structured options for merge, PR/MR, or cleanup

receiving-code-review

Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation

requesting-code-review

Use when completing tasks, implementing major features, or before merging/opening a PR/MR to verify work meets requirements

subagent-driven-development

Use when executing implementation plans with independent tasks in the current session

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

Testing & QA Listed

test-driven-development

Use when implementing any feature or bugfix, before writing implementation code

using-git-worktrees

Use when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback

verification-before-completion

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

writing-plans

Use when you have a spec or requirements for a multi-step task, before touching code

acli-jira

Atlassian CLI (acli) for Jira operations. Use when working with Jira issues via command line - viewing tickets ("view PROJ-123", "show ticket"), searching with JQL ("find issues", "search Jira"), listing projects/boards/sprints, viewing comments/attachments/links, or any Jira query that benefits from CLI output. Prefer acli over MCP tools for structured JSON output and complex JQL queries.

buffett-investment-strategy

Apply Warren Buffett's investment strategy framework derived from 60 years of Berkshire Hathaway shareholder letters (1965-2024) and 32 years of annual meeting transcripts (1994-2025). Use when: (1) evaluating a stock or business for investment, (2) analyzing competitive advantages or economic moats, (3) assessing management quality, (4) making capital allocation decisions, (5) reviewing portfolio strategy, (6) discussing value investing principles, (7) user mentions Buffett, Berkshire, value investing, moats, intrinsic value, or margin of safety.

deep-research

Deep multi-step web research using Agent Teams for parallel, coordinated investigation. Use when user needs comprehensive research on a topic ("deep research X", "research X thoroughly", "give me a full analysis of X"), wants a structured research report synthesizing multiple sources, or needs to investigate a topic from multiple angles. Triggers on phrases like "deep research", "thorough research", "comprehensive analysis", "research report", or "/deep-research".

Data & Documents Listed

facebook-scraper

Scrape posts (text, images, top-level comments) from a Facebook group or fan page and output a Markdown report. Use when: the user wants to scrape, back up, or collect FB group/page post content and comments. Trigger on: facebook group, fan page, post scrape, fb group/page scraper.

find-claude-session

Search across all past Claude Code sessions (across every working directory) to recall what was discussed and locate the original cwd. Use this skill whenever the user wants to find a past conversation, recall which directory a session was in, search their Claude history for a topic, or asks things like '我之前在哪個對話/目錄聊過 X', '幫我找之前討論 X 的 session', 'which directory did we talk about X in', 'search my Claude transcripts for Y'. Especially useful when the user opens throwaway conversations via `cd $(mktemp -d)` and later forgets which tmp directory hosted a discussion.

finmind

Query Taiwan stock financial reports via FinMind API. Use when: (1) analyzing Taiwan company financials (income statement, balance sheet, cash flow), (2) checking monthly revenue trends, (3) looking up dividend history or PER/PBR/yield, (4) comparing company financial metrics, (5) any question about Taiwan stock fundamentals. Trigger on stock codes like 2330, 2317, or company names like TSMC, Hon Hai, MediaTek.

gh-actions-permissions

Manage GitHub Actions workflow permissions using gh CLI. Use when: (1) Checking repository Actions permissions settings, (2) Enabling GitHub Actions to create or approve pull requests, (3) Troubleshooting workflow failures with "GitHub Actions is not permitted to create or approve pull requests" errors, (4) Configuring workflow permissions for tools like release-please that need PR creation access, or (5) Managing default_workflow_permissions and can_approve_pull_request_reviews settings.

glab-cli

GitLab CLI (glab) integration for managing repositories, merge requests, CI/CD pipelines, variables, and runners on self-managed GitLab instances. Use when working with GitLab - authentication issues ("token expired", "unauthorized"), repo operations ("create repo", "clone"), merge requests ("create MR", "view MR", "code review"), CI/CD pipelines ("check CI", "view job logs", "trigger job", "download artifacts"), variables ("list variables", "set variable"), or runners ("list runners").

go-concurrency-patterns

Master Go concurrency with goroutines, channels, sync primitives, and context. Use when building concurrent Go applications, implementing worker pools, or debugging race conditions.

malware-repo-analysis

Use when analyzing a third-party git repository for trojans, backdoors, data exfiltration, supply chain attacks, or other malicious code — NOT for finding code vulnerabilities or CVEs. Trigger on: "analyze this repo", "check for malware", "is this safe to use", "supply chain risk", "suspicious package".

ndc-lightscore

Fetch Taiwan NDC 景氣對策信號 latest month, score, signal light, recent series, full historical score series, consecutive red-light runs, and next publish date from index.ndc.gov.tw using a uv Python script with Patchright/Chrome when Cloudflare blocks direct HTTP requests.

DevOps & Infrastructure Listed

newcity

查詢 Newcity 社區物業 app(itlife.com.tw / NewcityWebApi)的信件包裹與社區點數。信件:當使用者想知道「有沒有我的信/包裹到了」「未領信件」「待領信件」「已領信件」「社區管理室有沒有東西要領」「幫我查 newcity / itlife 的信件」時使用。點數:當使用者想知道「我還有多少社區點數」「剩餘點數」「公設/餐飲點數還剩多少」「點數什麼時候過期」時使用。公告:當使用者想知道「社區有什麼新公告」「最新公告」「未讀公告」「管委會公告」「公告內容/附件」時使用。即使使用者沒明講 app 名稱,只要是查「社區/大樓管理室代收的信件包裹」「社區點數餘額」或社區公告就適用。本 skill 透過 app 的 API 重現登入->選社區->解析住戶->查詢流程,印出結果。

API & Backend Listed

postgres

Execute read-only SQL queries against multiple PostgreSQL databases. Use when: (1) querying PostgreSQL databases, (2) exploring database schemas/tables, (3) running SELECT queries for data analysis, (4) checking database contents. Supports multiple database connections with descriptions for intelligent auto-selection. Blocks all write operations (INSERT, UPDATE, DELETE, DROP, etc.) for safety.

powershell-style-guide

Review and write PowerShell code following community style guide and best practices (based on PoshCode/PowerShellPracticeAndStyle). Use when writing new PowerShell scripts, functions, or modules (.ps1, .psm1, .psd1), reviewing PowerShell code for style compliance, refactoring PowerShell code, or any task involving PowerShell scripting where code quality matters.

pre-goal

Use when the user is about to launch /goal (or any long-running autonomous agent task) and the task is still vaguely defined, mentions "pre-goal" / "define the goal" / "write a goal prompt", or complains that a previous /goal run drifted off course, wrapped up too early, or produced unsatisfying results. Guides the user to pin down the definition of done before launching.

prompt-engineering

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Data & Documents Listed

reimburse-request

Submit a company expense reimbursement (請款 / 報帳) on reimburse.digital using the agent-browser CLI. Use this whenever the user wants to claim an expense, file for reimbursement, or report a cost they paid for the company — wellness fund, health check-up, marriage/childbirth/funeral allowance, and other benefit claims. Trigger on phrases like "幫我請款", "我要報帳", "報銷", "核銷", "報支", "claim expense", "file a reimbursement", "submit an expense", or any request that names an amount plus what it was for. The skill logs in automatically (reading the organisation id and SSO credentials from a .env file), opens the Benefit/Others claim form, fills the details, attaches the receipt, and submits.

sap-leave-request

Submit a leave / time-off request in SAP SuccessFactors using the agent-browser CLI. Use this whenever the user wants to take leave, apply for vacation, or file any kind of absence — annual leave (特休), personal leave (事假), sick leave (病假), marriage/birthday/compassionate leave, work-from-home, etc. Trigger on phrases like "幫我請假", "我要請假", "SAP 請假", "要求休假", "請特休", "on leave", "apply for leave", "book time off", or any request that names a date range plus a leave type. The skill logs in automatically (reading credentials from a .env file), opens the 要求休假 form, picks the leave type, fills the dates, and submits.

shell-style-guide

Review shell/bash code for adherence to Google Shell Style Guide. Use when the user requests a code review of shell scripts (.sh, .bash), writing new shell scripts, fixing shell script issues, or checking shell code against style guidelines. Trigger phrases include: review this shell script, check shell style, write a bash script, review my shell code, or any task involving shell/bash scripting where code quality and consistency matter.

slack-notify

Send Slack notifications via Incoming Webhook with mrkdwn formatting. Use this skill whenever: (1) you need to send a message or notification to Slack, (2) alerting users about task completion or results, (3) sharing summaries, reports, or status updates to a Slack channel, (4) another skill or workflow needs to notify someone via Slack. Even if the user doesn't say 'Slack' explicitly, use this skill when they mention sending notifications, alerting a channel, or posting updates.

Web & Frontend Listed

web-browsing

Navigate, interact with, and read live web pages from the terminal by combining the agent-browser CLI (headless Chrome over CDP) with the defuddle CLI (clean Markdown extraction). Use whenever a task needs a real browser — opening a page, filling a form, clicking through a flow, logging in, taking a screenshot, testing a web app, scraping data — or whenever a URL needs to be read as clean Markdown instead of raw HTML, including JS-rendered SPAs and pages behind a login that a plain HTTP fetch cannot see. Triggers on "open a website", "read this page", "fetch this URL", "what does this page say", "fill out a form", "take a screenshot", "scrape this page", "log in to", "test this web app", 瀏覽器自動化, 開網頁, 讀這個網頁, 網頁截圖, 填表單, 抓網頁內容. Prefer this over WebFetch and over any other browser automation approach.

Data & Documents Listed

youtube-transcribe

Transcribe a YouTube video (or any local audio/video file) into a text transcript, SRT subtitles, and JSON using whisper locally. Use this whenever the user shares a YouTube URL and wants its subtitles, captions, transcript, 字幕, or 逐字稿 — especially when the video has no captions to download — or wants to transcribe / 轉錄 an audio or video file. Triggers on requests like "幫我抓這部 YouTube 的字幕/逐字稿", "transcribe this video", "把這段音檔轉成文字". Runs locally and is cross-platform: mlx-whisper on Apple Silicon, faster-whisper elsewhere.

gws-gmail

Operating Gmail through the `gws` (Google Workspace CLI) tool — reading, searching, labelling, archiving, deleting mail, and managing Gmail filters. Use this skill whenever a task involves triaging or reorganising a mailbox, building a label taxonomy, creating or editing Gmail filters, bulk-archiving or bulk-deleting messages, auditing what is cluttering an inbox, or any request phrased like "clean up my Gmail", "sort my inbox", "set up mail rules", "整理信箱", "建立郵件規則". Also use it whenever mail content will be summarised or acted on, because it carries the prompt-injection rules for untrusted message text. The Gmail API and the `gws` CLI have many sharp edges that silently produce wrong results — consult this skill before the first command, not after something breaks.

chrome-devtools

Use when about to use chrome-devtools MCP tools and Chrome connection fails or remote debugging needs to be enabled. Symptoms include "Could not connect to Chrome" or "Could not find DevToolsActivePort" errors.

grilling

Interview the user relentlessly about a plan or design. Use when the user wants to stress-test a plan before building, or uses any grill trigger phrase.

Web & Frontend Listed

styling

CSS and Tailwind styling guidelines. Use when writing styles, creating UI components, reviewing CSS/Tailwind code, or deciding on wrapper element structure.