黎子健
Community@Junkfooooood · 中国
Junkfooooood's Hermes skill registry spans LLM fine-tuning and serving, agentic coding workflows, macOS productivity, and generative media creation.
Agent Skills by 黎子健
Showing 82 vetted skills indexed across 1 GitHub repositories.
Junkfoooooodyuanbao
Mention users, query group info, and send direct messages in Yuanbao groups.
Junkfooooooddogfood
Tests web applications through browser automation and generates structured bug reports with screenshots.
Junkfoooooodapple-reminders
Manage Apple Reminders from the terminal using the remindctl command-line tool.
Junkfoooooodimessage
Send and read iMessage and SMS messages via the imsg CLI on macOS.
Junkfoooooodfindmy
Track Apple devices and AirTags via FindMy.app UI automation on macOS.
Junkfoooooodapple-notes
Create, search, edit, and export Apple Notes via the memo CLI on macOS.
Junkfoooooodblogwatcher
Monitor blogs and RSS/Atom feeds using the blogwatcher-cli command-line tool.
Junkfoooooodresearch-intelligence
Collects, triages, cross-verifies, and synthesizes external sources into evidence-backed research briefs.
Junkfoooooodpolymarket
Query Polymarket prediction markets for prices, orderbooks, and price history via public REST APIs.
Junkfoooooodllm-wiki
Build and maintain an interlinked markdown knowledge base from ingested sources.
Junkfoooooodarxiv
Search and retrieve academic papers from arXiv and Semantic Scholar APIs.
Junkfoooooodresearch-paper-writing
Guides end-to-end ML research paper production from experiment design through conference submission.
Junkfoooooodminecraft-modpack-server
Configures and launches modded Minecraft servers from CurseForge and Modrinth server packs.
Junkfoooooodpokemon-player
Plays Pokemon games via headless PyBoy emulation with RAM state reads and screenshot verification.
Junkfoooooodxurl
Post, search, and manage X/Twitter content via the official xurl CLI.
Junkfoooooodwebhook-subscriptions
Create webhook subscriptions that trigger agent runs from external service events.
Junkfoooooodjupyter-live-kernel
Execute stateful Python code through a live Jupyter kernel with persistent variables.
Junkfoooooodtest-driven-development
Enforces red-green-refactor TDD workflow requiring failing tests before production code.
Junkfoooooodsystematic-debugging
Diagnose software bugs through a four-phase root cause investigation workflow.
Junkfoooooodplan
Writes actionable markdown implementation plans to .hermes/plans/ without executing code.
Junkfoooooodhermes-agent-skill-authoring
Author and validate in-repo SKILL.md files with correct frontmatter and structure.
Junkfoooooodproject-architecture-documentation
Generate Markdown architecture documentation from existing project files, configs, and code comments.
Junkfoooooodpython-debugpy
Debug Python code with pdb breakpoints, post-mortem inspection, and debugpy remote attach.
Junkfoooooodwriting-plans
Write implementation plans with bite-sized tasks, exact file paths, and complete code.
Frequently Asked Questions About 黎子健
FAQPage SchemaWhat tasks can I accomplish with Junkfooooood's skill registry?▼
You can fine-tune and serve LLMs (TRL, Axolotl, Unsloth, vLLM, llama.cpp), benchmark models with lm-evaluation-harness, manage GitHub repos and PRs, debug Python/Node code, extract PDF text, control macOS apps like Reminders and iMessage, and generate music, pixel art, diagrams, and infographics.
Who is the target audience for these skills?▼
ML engineers running post-training pipelines (SFT, DPO, RLHF), software developers using agentic coding delegation via Claude Code, Codex, or OpenCode, researchers writing NeurIPS/ICML papers, and power users automating macOS, Notion, Linear, Airtable, and Google Workspace environments.
How are these skills installed and run?▼
Skills are SKILL.md packages for the Hermes Agent runtime, invoked through its CLI or gateway. Many require prerequisite binaries (remindctl, imsg, himalaya, pygount, xurl) or Python dependencies (torch, transformers, trl, vllm), and some need API keys such as LINEAR_API_KEY, NOTION_API_KEY, or TENOR_API_KEY.
Are these skills open source and free to use?▼
Nearly all skills carry the MIT license, including the fine-tuning, debugging, GitHub, and creative suites. The sole exception is the powerpoint skill, which is proprietary with terms defined in its bundled LICENSE.txt file.
What platforms and dependencies are required?▼
Apple ecosystem skills (Reminders, iMessage, FindMy, Notes) require macOS. ML skills require PyTorch with CUDA-capable GPUs for training, while llama.cpp supports CPU and Apple Silicon inference. Linux and macOS are supported for research-paper-writing and most CLI-based skills.