By snkrheadz
Research pack: arxiv-ai-researcher (AI/ML paper discovery & synthesis), gemini-api-researcher (Gemini API capabilities & usage), and huggingface-spaces-researcher (HF Spaces/model discovery). Role-agnostic research tooling for anyone investigating AI/ML papers, APIs, and models.
Use this agent when you need to research AI models, architectures, or techniques based on academic literature from arXiv. This includes understanding new model papers, comparing different approaches, summarizing research findings, extracting implementation details from papers, or staying updated on state-of-the-art developments in machine learning and artificial intelligence. Examples: <example> Context: User wants to understand a specific AI model architecture. user: "Can you explain how the Mamba architecture works and how it compares to Transformers?" assistant: "I'll use the arxiv-ai-researcher agent to research the Mamba architecture from relevant arXiv papers and provide a comprehensive comparison with Transformers." <commentary> Since the user is asking about a specific AI architecture that has been published on arXiv, use the arxiv-ai-researcher agent to find and synthesize information from relevant academic papers. </commentary> </example> <example> Context: User needs to find state-of-the-art approaches for a specific AI task. user: "What are the latest approaches for efficient LLM inference?" assistant: "Let me launch the arxiv-ai-researcher agent to survey recent arXiv publications on efficient LLM inference techniques and summarize the key findings." <commentary> Since the user is asking about recent developments in AI research, use the arxiv-ai-researcher agent to search and synthesize relevant papers from arXiv. </commentary> </example> <example> Context: User wants implementation details from a research paper. user: "I found a paper about FlashAttention. Can you help me understand the key implementation details?" assistant: "I'll use the arxiv-ai-researcher agent to analyze the FlashAttention paper and extract the key implementation details, algorithms, and optimization techniques." <commentary> Since the user needs detailed technical information from an academic paper, use the arxiv-ai-researcher agent to thoroughly analyze the paper and extract relevant implementation details. </commentary> </example>
Google Gemini API research and implementation support agent. Specializes in Gemini API via ai.google.dev. Use gcp-best-practices-advisor for Vertex AI routes. Trigger examples: - How to use Gemini API - How to use generateContent - Multimodal processing with Gemini - Function Calling implementation - Gemini pricing information - Gemini 2.0 Flash new features - google-genai SDK usage - Real-time processing with Live API
Use this agent when you need to research, analyze, or compare HuggingFace Spaces for AI/ML demos and prototypes. This includes finding implementation examples, understanding how models are deployed as interactive demos, extracting code patterns from Spaces, or discovering research paper demonstrations. Examples: <example> Context: User wants to find demos related to a specific AI technique. user: "Find me Spaces that demonstrate Stable Diffusion XL" assistant: "I'll launch the huggingface-spaces-researcher agent to search for SDXL-related Spaces and analyze their implementations." <commentary> Since the user is looking for specific model demos, use the huggingface-spaces-researcher agent to search and analyze relevant Spaces. </commentary> </example> <example> Context: User needs to understand implementation patterns from HuggingFace Spaces. user: "How do popular Gradio apps handle image generation queues?" assistant: "Let me use the huggingface-spaces-researcher agent to analyze several popular image generation Spaces and extract their queue handling patterns." <commentary> Since the user needs implementation insights from multiple Spaces, use the huggingface-spaces-researcher agent for comprehensive analysis. </commentary> </example> <example> Context: User wants to find the official demo for a research paper. user: "Is there a demo for the Segment Anything paper?" assistant: "I'll use the huggingface-spaces-researcher agent to find official and community demos for the Segment Anything Model (SAM)." <commentary> Since the user is looking for paper-related demos, use the huggingface-spaces-researcher agent to search and verify official implementations. </commentary> </example>
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code skill marketplace maintained by snkrheadz. Goal: a new teammate (engineer / PM) can use Claude Code in the same environment from day one.
We intentionally keep this lean. Anything the official Claude Code commands already cover (code review, simplification, app verification, committing, project init) is not duplicated here — see Covered by official Claude Code. This repo only ships the gaps.
A plugin alone is not enough. By design, Claude Code cannot distribute
permissions.deny (the security deny-list) or the global CLAUDE.md (your working
philosophy) through a plugin, so those go through a separate channel.
| Channel | What | How |
|---|---|---|
| A. Plugins | skills / agents / hooks | /plugin marketplace add → /plugin install |
| B. Settings + philosophy | permission deny-list / CLAUDE.md | paste into ~/.claude/settings.json + install CLAUDE.md |
/plugin marketplace add snkrheadz/claude-skills
Works for private repos too. A manual
addreuses your local git auth (gh auth login/ SSH / Keychain), so if you cangit cloneit, you can add it. Only setGITHUB_TOKEN(withreposcope) if you also want auto-update on startup.
/plugin install core@claude-skills
What Core ships (role-agnostic):
first-principles — rethink a problem from fundamentalsteach-session — teach back the work just done (great for onboarding)html-output — emit specs / reviews / reports as rich HTMLpre-tool-guard hook — block access to sensitive files (defense in depth)# Our own business assets (maintained by snkrheadz)
/plugin install pm@claude-skills # 業務定義シート (task definition sheet) as A4 HTML
# PM lifecycle (external, MIT, recommended for non-developers)
/plugin marketplace add phuryn/pm-skills
/plugin install pm-product-discovery@pm-skills # discovery / prioritization / interviews
/plugin install pm-product-strategy@pm-skills # strategy / canvases / pricing
/plugin install pm-execution@pm-skills # PRD / OKR / roadmap / sprint
/plugin install pm-market-research@pm-skills # personas / market sizing / competitors
# Install only what you need. All 9 plugins: https://github.com/phuryn/pm-skills
# Document generation (official, docx/pptx/xlsx/pdf)
/plugin install document-skills@anthropic-agent-skills
/plugin install eng@claude-skills
This pack deliberately does not re-implement review/simplify/verify/commit — those are official commands now (see below). It ships the workflow gaps around them:
merge-pr review-inbox test-and-fix refactor-swarm techdebt
trace-dataflow db-querycode-architect architecture-reviewer verify-shell
migration-assistant oncall-guide state-machine-diagram
aws-best-practices-advisor gcp-best-practices-advisorRecommended alongside the official LSP plugins:
/plugin install gopls-lsp@claude-plugins-official # Go
/plugin install typescript-lsp@claude-plugins-official # TS
For anyone investigating AI/ML papers, APIs, and models — any role can add it.
/plugin install research@claude-skills
Agents (3): arxiv-ai-researcher (paper discovery & synthesis),
gemini-api-researcher (Gemini API capabilities & usage),
huggingface-spaces-researcher (HF Spaces / model discovery).
These roles are served almost entirely by official Anthropic plugins — we no longer ship a custom pack for them.
# Marketer
/plugin install document-skills@anthropic-agent-skills # brand-guidelines, internal-comms, doc-coauthoring
# + the deep-research skill; landing pages via frontend-design@claude-plugins-official
# Designer
/plugin install frontend-design@claude-plugins-official
/plugin install document-skills@anthropic-agent-skills # canvas-design, theme-factory, algorithmic-art, slack-gif-creator
These workflows ship with Claude Code itself, so this marketplace intentionally omits them. Use the official commands directly:
| Need | Official command |
|---|---|
| Review the current diff / a PR | /code-review (/code-review ultra [PR#] for deep multi-agent cloud review), /review |
| Security review | /security-review |
| Simplify / de-duplicate code | /simplify |
| Verify a change by running the app | /verify, /run |
| Commit | Claude Code commits natively (or the commit-commands plugin) |
Initialize CLAUDE.md / project config | /init, the update-config skill |
| Claude Code / Agent SDK / API how-to | the official claude-code-guide agent |
npx claudepluginhub snkrheadz/claude-skills --plugin researchEngineering workflow pack (fills gaps the official commands don't cover): merge-pr, review-inbox, test-and-fix, refactor-swarm, techdebt, trace-dataflow, db-query — plus code-architect, architecture-reviewer, verify-shell, migration, oncall and AWS/GCP advisor agents.
Cross-role foundation: first-principles thinking, teaching/onboarding, rich HTML output, and a sensitive-file guard hook. Role-agnostic — install first.
PM / business assets maintained by snkrheadz: 業務定義シート (task definition sheet) as A4 HTML. Pair with phuryn/pm-skills for the full product lifecycle.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Agent that simplifies and refines code for clarity, consistency, and maintainability while preserving functionality
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Curate auto-memory, promote learnings to CLAUDE.md and rules, extract proven patterns into reusable skills.