メニュー
Guide for creating effective skills that extend different agents' capabilities. Use when creating new skills or updating existing skills with specialized knowledge, workflows, or tool integrations.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SOC 職業分類に基づく
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/susamn/aistuff --skill skill-creatorコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
SKILL.md を表示中
Manage the dotfiles repository, stow packages, and configure agent skills. Use when adding a skill, modifying the stow setup, or working inside ~/dotfiles.
Practice LeetCode problems in a 3-tab mosaic dashboard (Problem, Intuition, Solution — Python + Go). Starter-seeded with 100 classic problems; add, swap, or remove any problem by URL or description. Use when asked to practice LeetCode, add/swap a coding-interview problem, or open the LeetCode practice dashboard.
Import external skills from Git repositories, local directories, or web URLs into the local skills library, adapting them to house conventions, auditing, and suggesting cross-skill integrations. Use when importing a skill, bringing in an external skill, or adding a skill from GitHub.
| name | skill-creator |
| description | Guide for creating effective skills that extend different agents' capabilities. Use when creating new skills or updating existing skills with specialized knowledge, workflows, or tool integrations. |
| version | 2.4.0 |
| kind | hybrid |
| triggers | ["create a new skill","add a skill"] |
| intent | meta |
| config_dir | ~/.config/skill-config/skill-creator |
| guardrails | ["Classify the skill before writing it. Guidance skills get zero scripts; pipeline skills keep judgment out of code.","A script must never dump raw data into agent context — emit a projection plus a handle to the full artifact.","SKILL.md is a router. Detail belongs in references/, loaded on demand.","A skill update that changes behavior updates its prompts/ too, if it has one — stale examples are worse than none."] |
| resources | ["<SKILL_PATH>/scripts/scaffold.sh","<SKILL_PATH>/references/frontmatter.md","<SKILL_PATH>/references/script-contract.md","<SKILL_PATH>/references/guidance-skills.md","<SKILL_PATH>/references/data-app-skills.md","<SKILL_PATH>/references/webapp-list-view.md","<SKILL_PATH>/references/webapp-card-board.md","<SKILL_PATH>/prompts/create-data-app-skill.md","<SKILL_PATH>/prompts/migrate-to-data-app-skill.md"] |
| tools | ["bash"] |
| created_at | 2026-05-30T00:00:00.000Z |
| updated_at | 2026-07-30T00:00:00.000Z |
Read references/house-conventions.md first. It carries the path, tooling,
deployment, and working-style conventions of this setup — a skill that ignores them
is valid and still feels foreign. Optional config:
~/.config/skill-config/skill-creator/skill.properties (default_author,
auto_stow_after_creation), protocol in references/frontmatter.md.
Every skill is one of three kinds, declared as kind: in frontmatter. The kind
determines the entire shape of the skill. Choosing wrong is the most expensive
mistake available here, because it is discovered only after the skill is written.
| kind | test | shape |
|---|---|---|
guidance | Encodes judgment. Nothing to compute at runtime. | Prose only. Zero scripts. Router SKILL.md + references/. |
pipeline | Operates on real data — files, repos, APIs, metrics. | Thin SKILL.md + scripts/ + a declared output contract. |
hybrid | Both, in cleanly separable sections. | Declare which sections are which. Suspect it is really two skills. |
If you cannot name the concrete runtime input a script would read, it is
guidance. Adding scripts to a guidance skill produces rigid output that is
wrong in every situation the author did not foresee.
Do not ask "can this be scripted." Ask whether the operation is total or a judgment.
Corollary: tunable values live in data, the logic applying them lives in code. Thresholds, weights, and limits belong in a JSON or properties file beside the script, never inline. Retuning must be an edit to data, not to logic.
A pipeline skill has two places data crosses a line. Both must be designed; only the first is obvious.
Compression is not free — it is inference relocated to authoring time, where you know less than the agent will at runtime. So compress to preserve decision-relevant variance, and always leave the escape hatch. A summary with no handle back to the data is a dead end.
Full contract: references/script-contract.md. Read it before writing any
pipeline skill.
One script with modes beats many single-purpose scripts. Every script is a
contract to keep in sync, so N scripts means N places for schema drift — prefer
run.sh analyze | summary | verbose over three separate files.
Two counterweights, both real failure modes:
references/<topic>.md rather than
trimming words. An oversized SKILL.md with no references/ is the actual failure
this guards against.kind. State it and the reasoning to the user.pipeline, name the exact
runtime input and the decision its output supports. Also ask: does this need
a live dashboard on mosaic, or is a generated report/artifact enough? If yes,
it's data-app-backed — references/data-app-skills.md.<SKILL_PATH>/scripts/scaffold.sh <name> --kind <kind> [--webapp]<SKILL_PATH>/scripts/<script>; use $TOOLS_PATH / $SCRIPTS_PATH only for
genuinely global utilities living outside the skill.<SKILL_PATH>/prompts/<topic>.md now, while the skill's
purpose and triggers are freshest — this is the best time, not an
afterthought. Short, directly usable example prompts, one file per distinct
usage scenario. Skip it only when triggers: alone already says everything
there is to say; write it once usage has more than one shape (e.g. "create
new" vs. "migrate existing") or involves a reusable template worth saving —
see skill-creator's own prompts/ for what that looks like.## Available skills table in
~/dotfiles/workspace/aistuff/skills/AGENTS-TEMPLATE.md.bash ~/dotfiles/do-stow.sh~/dotfiles/workspace/aistuff/skills/skill-manager/scripts/audit.sh <name>.
Must pass before committing.workspace/aistuff/ is a git submodule, so this is two commits:
git -C ~/dotfiles/workspace/aistuff add skills/<name>/ skills/AGENTS-TEMPLATE.md
git -C ~/dotfiles/workspace/aistuff commit -m "skills: add <name>"
git -C ~/dotfiles add workspace/aistuff
git -C ~/dotfiles commit -m "skills: bump submodule for <name>"
prompts/ exists, revisit it in the same pass. Stale
examples that no longer match current behavior are worse than no examples
at all — they get trusted and then fail.Rename the directory to <name>.disabled and re-run do-stow.sh — the suffix is
the only mechanism. Update the table's Enabled cell in the same change, or the
two sources drift. Detail: dotfiles-management.
config_dir is
optional, declared only when the skill has real persistent state| file | when |
|---|---|
references/house-conventions.md | always — before writing or editing any skill |
references/frontmatter.md | writing frontmatter; the config protocol |
references/script-contract.md | any pipeline skill — required |
references/guidance-skills.md | any guidance skill; splitting an oversized skill |
references/data-app-skills.md | skill needs a live mosaic dashboard, not just a report |
references/webapp-list-view.md | data-app's main view is a manifest-driven list with drill-down |
references/webapp-card-board.md | data-app is a single-page Trello/Keep-style card board |
prompts/create-data-app-skill.md | starting prompts for a brand-new data-app skill |
prompts/migrate-to-data-app-skill.md | reusable template for porting an existing skill onto mosaic |