dotfiles

user invoke only

user invoke only

Official Claude Code reference for agent skills. Use this when configuring frontmatter or other settings for agent skills.

user invoke only

Use this to consult the official Pi coding agent reference when investigating, designing, or implementing Pi — setup, configuration, providers/models, skills, extensions, packages, prompt templates, sessions, SDK, RPC, JSON event stream, and terminal setup.

trigger is `visualize`

A skill where the LLM manages the Wiki of an Obsidian Vault (memos). Always use this skill in situations such as: - 「WikiにXXを取り込んで」「ingestして」「このメモをWikiに追加して」「Wikiを更新して」 - 「Wikiで調べて」「Wikiを検索」「WikiでXXについて教えて」 - 「Wiki lint」「Wikiの健全チェック」「stale な refs を確認」「orphan pages を探して」 - Activate it proactively when referencing or updating the Wiki is appropriate in the flow of conversation

user invoke only

Extract lessons that make next time easier. Fire proactively when: (1) a reusable insight emerges at a work boundary — a bug workaround found, a gotcha cleared, the right procedure/command/config figured out after trial and error, a tip for avoiding a repeated mistake. (2) a session wraps up — review the whole session and extract its lessons.

An orchestration skill that collects references from documentation URLs on the web and chains into skill creation (skill-creator). Always activate it for requests like 「URLからスキルを作って」「このドキュメントを元にスキル化したい」「firecrawlでリファレンスを集めてスキルを作って」.

user invoke only

GitHub issue を sub-issue に分解し、worktree 分離した並行エージェントで実装、読み取り専用の検証ゲートを経て、コンフリクト検知とマージ順序決定まで行う並行開発オーケストレーション手順。/parallel-issues <issue番号 or タスク概要> で起動する。

Raindrop.io REST API reference. Use for bookmarks (raindrops), collections, highlights, tags, filters, user, import/export/backups endpoints, OAuth authentication, or building Raindrop.io integrations/bots.

Harvest un-distilled reusable knowledge from agent memory stores (Claude Code auto-memory, Serena memories) and promote it into the learning tiers (global CLAUDE.md / skill / <repo>/.learnings/), then dispose of the source entries to avoid dual management. Use when the user runs /learning-from-memory, or explicitly asks to 「メモリを蒸留して」 「auto-memory から学びを昇格して」「メモリの棚卸しをして」. Do NOT fire proactively at work boundaries or session wrap-up — that is the learning skill's job.

Conventions for the file format and structure of documents (Markdown, etc.). Always use this skill when creating, editing, or formatting Markdown files such as documents, READMEs, design docs, research notes, or procedure guides.

Delegate small, well-scoped coding tasks to the local pi worker through Ollama. Use when Codex needs a secondary pass for lightweight implementation work such as fixing pyright, type-checker, linter, or warning output; writing focused tests to improve coverage; asking for a minimal patch plan; or getting a second implementation suggestion for a narrow coding task.

A skill for designing Claude Code agent team configurations. Use it proactively in situations such as: - When agent teams or multi-agent setups come up - When you want multi-angle verification such as architecture review, domain-expert review, or testing - When splitting and coordinating a large or complex task across multiple sessions - When asked how to use TeamCreate / SendMessage / TaskCreate Use it even when "agent team" is not explicitly stated, whenever quality verification or role division is being discussed.

Comprehensive official reference for Hermes Agent. Whenever the user asks about or requests building Hermes Agent deployment, configuration, or Slack integration, always activate this skill and read the appropriate file under references/.

Writing standards for Japanese technical documents and book manuscripts. Use when writing or revising/rewriting chapters, drafts, articles, or explanatory text for Japanese technical books.

Uses Jules (an asynchronous AI agent) to automate and manage large-scale codebase changes, test additions, refactoring, and dependency updates. Always use this skill when the user mentions 「大規模な修正を依頼したい」「時間がかかる作業をバックグラウンドでやってほしい」「Julesを使って」, or when a refactoring is too broad to do manually (e.g., project-wide symbol renaming, adding unit tests across all files).

A skill for applying the defined coding rules to Kotlin coding. Always activate this skill and apply the rules when writing Kotlin code or when asked to modify .kt files.

Automatically organizes an Obsidian vault. Use for requests such as "Obsidian整理", "vault整理", "ダッシュボード整理", "inbox整理", "メモ整理".

A skill that delegates code review and fixes to a subagent. Use when instructed with 「セルフコードレビューして」「サブエージェントにレビューさせて」.

A skill for applying the defined coding rules to Swift coding. Always activate this skill and apply the rules when writing Swift code or when asked to modify .swift files.

A TDD-style procedure for safely carrying out refactoring. Always propose this procedure before starting code refactoring, structural improvement, internal cleanup, renaming, or deduplication. Applies to any change that improves internals without altering external behavior.

user invoke only

user invoke only

user invoke only

user invoke only

Run any question, idea, or decision through a council of 5 AI advisors who independently analyze it, peer-review each other anonymously, and synthesize a final verdict. Based on Karpathy's LLM Council methodology. MANDATORY TRIGGERS: 'council this', 'run the council', 'war room this', 'pressure-test this', 'stress-test this', 'debate this'. STRONG TRIGGERS (use when combined with a real decision or tradeoff): 'should I X or Y', 'which option', 'what would you do', 'is this the right move', 'validate this', 'get multiple perspectives', 'I can't decide', 'I'm torn between'. Do NOT trigger on simple yes/no questions, factual lookups, or casual 'should I' without a meaningful tradeoff (e.g. 'should I use markdown' is not a council question). DO trigger when the user presents a genuine decision with stakes, multiple options, and context that suggests they want it pressure-tested from multiple angles.

user invoke only

user invoke only

user invoke only

user invoke only

user invoke only

user invoke only

Enforces the use of a standard `default` task in Taskfiles that lists all available tasks. Use this skill whenever creating, modifying, or auditing a `Taskfile.yml` or `taskfile.yml`.

Alpaca Markets API reference. Use for trading/market data/broker API, orders, positions, account management, or building Alpaca-based bots/apps.

Bluesky (AT Protocol) API reference. Use for endpoints, XRPC methods, app.bsky*/chat.bsky* APIs, or building Bluesky bots/clients/integrations.

llama.cpp setup expert — Mac (Apple Silicon/Intel) Metal GPU acceleration, memory optimization (KV quantization, Flash Attention), multimodal models (LLaVA/Gemma3/Qwen-VL), and server configuration.