Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
Skills(SKILL.md)は、AIエージェント(Claude Code、Cursor、Codexなど)に特定の能力を追加するための設定ファイルです。
詳しく見る →Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
Evaluates NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments. Use when setting up cosmos-policy for robot manipulation evaluation, running headless GPU evaluations with EGL rendering, or profiling inference latency on cluster or local GPU machines.
Guides researchers through structured ideation frameworks to discover high-impact research directions. Use when exploring new problem spaces, pivoting between projects, or seeking novel angles on existing work.
Applies cognitive science frameworks for creative thinking to CS and AI research ideation. Use when seeking genuinely novel research directions by leveraging combinatorial creativity, analogical reasoning, constraint manipulation, and other empirically grounded creative strategies.
Workflow orchestrator that chains existing skills for feature development
Modular Code Organization
Fast file editing via Morph Apply API (10,500 tokens/sec, 98% accuracy)
Parallel Agent Type Contracts
Identify failure modes before they occur using structured risk analysis
Create and use Claude Code slash commands - quick prompts, bash execution, file references
Wiring Verification
Assess whether your product work is AI-first or AI-shaped. Use when evaluating AI maturity and choosing the next team capability to build.
Understand the PM-to-Director transition through altitude and horizon thinking. Use when diagnosing scope, time-horizon, or leadership-level gaps.
Diagnose context stuffing vs. context engineering. Use when an AI workflow feels bloated, brittle, or hard to steer reliably.
Plan customer discovery interviews with the right goal, segment, constraints, and method. Use when preparing interviews for problem validation, churn research, or new product ideas.
Run a full discovery cycle from problem hypothesis to validated solution. Use when a team needs a structured path through framing, interviews, synthesis, and experiments.
Plan a VP or CPO 30-60-90 day diagnostic onboarding path. Use when entering a new executive product role and avoiding premature change.
Create a Geoffrey Moore-style positioning statement. Use when clarifying who you serve, what problem you solve, your category, and why you're different from alternatives.
Run a positioning workshop that surfaces target customer, unmet need, category, benefits, and differentiation. Use when your product messaging feels fuzzy, generic, or misaligned.
Build a structured PRD that connects problem, users, solution, and success criteria. Use when turning discovery notes into an engineering-ready document for a major initiative.
Choose a prioritization framework based on stage, team context, and stakeholder needs. Use when deciding between RICE, ICE, value/effort, or another scoring approach.
Guide teams through MITRE's Problem Framing Canvas. Use when you need a clearer problem statement before jumping to solutions.
Run an end-to-end product strategy session across positioning, discovery, and roadmap planning. Use when a team needs validated direction before committing to execution.
Evaluate an AI product idea across outcomes, hypotheses, risks, and positioning. Use when deciding whether an AI solution deserves investment or recommendation.
Plan a strategic roadmap across prioritization, epic definition, stakeholder alignment, and sequencing. Use when turning strategy into a release plan that teams can execute.
Run a user story mapping workshop with adaptive questions and a structured map output. Use when you need backbone activities, tasks, and release slices for a workflow.
Automate Basecamp project management, to-dos, messages, people, and to-do list organization via Rube MCP (Composio). Always search tools first for current schemas.
Automate Coda tasks via Rube MCP (Composio): manage docs, pages, tables, rows, formulas, permissions, and publishing. Always search tools first for current schemas.
Automate Freshdesk helpdesk operations including tickets, contacts, companies, notes, and replies via Rube MCP (Composio). Always search tools first for current schemas.
Automate Gmail tasks via Rube MCP (Composio): send/reply, search, labels, drafts, attachments. Always search tools first for current schemas.
Automate Google Analytics tasks via Rube MCP (Composio): run reports, list accounts/properties, funnels, pivots, key events. Always search tools first for current schemas.
Create and edit Obsidian Bases (.base files) with views, filters, formulas, and summaries. Use when working with .base files, creating database-like views of notes, or when the user mentions Bases, table views, card views, filters, or formulas in Obsidian.
Automate Outlook Calendar tasks via Rube MCP (Composio): create events, manage attendees, find meeting times, and handle invitations. Always search tools first for current schemas.
Automate Render tasks via Rube MCP (Composio): services, deployments, projects. Always search tools first for current schemas.
Automate Salesforce tasks via Rube MCP (Composio): leads, contacts, accounts, opportunities, SOQL queries. Always search tools first for current schemas.
Automate Slack messaging, channel management, search, reactions, and threads via Rube MCP (Composio). Send messages, search conversations, manage channels/users, and react to messages programmatically.
Automate Stripe tasks via Rube MCP (Composio): customers, charges, subscriptions, invoices, products, refunds. Always search tools first for current schemas.
Automate Trello boards, cards, and workflows via Rube MCP (Composio). Create cards, manage lists, assign members, and search across boards programmatically.
Automate Twitter/X tasks via Rube MCP (Composio): posts, search, users, bookmarks, lists, media. Always search tools first for current schemas.
Automate Vercel tasks via Rube MCP (Composio): manage deployments, domains, DNS, env vars, projects, and teams. Always search tools first for current schemas.
Automate Zoom meeting creation, management, recordings, webinars, and participant tracking via Rube MCP (Composio). Always search tools first for current schemas.
Interactive setup wizard for the claude-ops plugin. Installs missing CLIs, configures env vars for each channel (Telegram, WhatsApp, Email, Slack, Notion, Linear, Sentry, Vercel), builds the project registry, and saves user preferences. Run once after installing the plugin or any time to reconfigure.
>
Archive completed milestone and prepare for next version
Route freeform text to the right GSD command automatically
Initialize a new project with deep context gathering and PROJECT.md
Check project progress, show context, and route to next action (execute or plan)
Execute a quick task with GSD guarantees (atomic commits, state tracking) but skip optional agents
Retroactively verify threat mitigations for a completed phase
Look up a vulnerability by ID or list all vulnerabilities for a package