All authors
Claude Skills by shimo4228
github.com/shimo422885 skills0 installs86 views
- Writing Ecosystem人間向け記事・エッセイ・ブログポスト・ニュースレターの唯一の執筆 orchestrator。project の publication channel contract を読み、中心命題 1 つの editorial brief、因果線、証拠の選択と除外、構成、執筆、review panel、著者の内容 GO、title-reviewer、quality-gate までを統括する。Use when — 「この記事を書いて」「このテーマでエッセイにして」「原稿の論点を一つに絞って構造改稿して」のような新規執筆・全体改稿・全文の別 channel 展開。NOT for — 一文や段落だけの翻訳（→ prose-translation）、title だけ（→ headline-craft / title-reviewer）、SNS 下書き（→ x-draft）、公開 thread 返信（→ public-comment）、AI 向け docs、README、paper、媒体固有の公開操作。Votes: 0GitHub stars: 3
- Zenn FormatZenn記事のfrontmatter、emoji/topics、Zenn固有Markdown記法の正本。Use when — Zenn原稿を作成・検証するとき、本文凍結後にtopics/emoji候補を提示するとき。NOT for — 執筆構成・voice・タイトル判定・公開可否・Dev.to/note/Substack形式。Votes: 0GitHub stars: 3
- Jev Judgment DesignHow to hand over judgments and decide accept/reject when moving closed judgments an LLM used to make (is this source relevant, is it new, how strong is the evidence) to TypeSafe's Jev. Putting what the judgment is made against into state, the bottom tier of a Score, Jev answers and code decides, the unit of judgment, sources that must pass (canary), calling from Pydantic AI. Use when — "I want to replace the LLM's judgment with Jev", "build a relevance judgment with Jev", "Jev's judgment lets...Votes: 0GitHub stars: 3
- Author Calibrated EvalHow to build and run a loop that polishes LLM-written reading material (body text, summaries, explanatory prose) by treating the author's reading as ground truth. Fix the materials, build a hard-case set, run blind side-by-side reads by the author, measure the ceiling with a strong model's reference draft, and keep the LLM judge dedicated to a faithfulness cutoff. Invoke with /author-calibrated-eval. NOT for — designing the judge's prompt or verdict format (→ llm-as-judge), validity of measur...Votes: 0GitHub stars: 3
- Review When Watchlaunchd が毎朝 06:10（jev-research-pipeline の jrp run の後）に起動する job。jrp が vault の daily-research/ に書いた新しいノート 1 本ごとに、この harness の RFC（frontmatter の review-when:）と ADR（## Review-when 節）に書かれた見直し条件を TypeSafe Jev に 1 組ずつ聞き、条件を満たしたノートを Slack に送る。model からは呼ばない — この文書は運用手引き。Votes: 0GitHub stars: 3
- Dr Reviewdaily-research の自己改善レビュー (ADR-0006)。metrics.jsonl の運用トレンド・決定論 lint・wiki 品質注記・DR-Expect 突合を集計し、改善候補を人間の判断材料として提示する。ユーザーが「dr review」「パイプラインの調子を見て」「daily-research の改善レビュー」と言ったとき、または /dr-review で起動。週 1 目安。判断と改善の着手はユーザーが行う — この skill は材料整形まで。Votes: 0GitHub stars: 3
- daily-researchAutomated daily per-repo research — cron-driven pipeline that runs claude -p inside each mapped research repo, explains one external development relevant to the repo and what it means for it, verifies the claims against their sources, and writes free-form explanatory research notes to an Obsidian vault. Bash orchestration with prompts over claude -p; JSON/TOML parsing in stdlib python3 (no pip dependencies).Votes: 0GitHub stars: 3
- Mono Figure記事や README の図（線・矢印・数字・ラベルで仕組みを見せる静的な図）を、mono-color の見た目（紙・インク・書体・余白）で作る橋渡し。既定は Claude が SVG で描く（Zenn などは PNG に撮る）。画像生成で作るときは、mono-color が組んだ prompt を Herdr で立てた Codex に渡す。Use when 「この節の図を mono-color で描いて」「README の概要図を mono-color の見た目で作って」「この図を Codex で画像生成してみて」。NOT for — 文字や数字の無いポスター・カバー（→ mono-color をそのまま）、操作できる構成図（→ archify）、データのグラフ（→ dataviz）、図を置くか・何枚か・置き場の約束（→ その repo の規約。Zenn は zenn-format の Figures、README は readme-writer の Visual）。Votes: 0GitHub stars: 3
- Read Only InstrumentsDesign pattern for read-only instruments — aggregate readings over an agent's stored state (distributions, compositions, cluster structure) that inform the operator before an intervention. Use when quantifying an open observation before intervening (instrument-first sequencing), when a design floor/threshold would otherwise be guessed, when calibrating an embedding-based reading (three-point scale), or when deciding whether to build OR remove an instrument (signal-first both ways). NOT for pe...Votes: 0GitHub stars: 2
- Replayable Audit LogsDesign pattern for observability-by-default in LLM agents — every feature that performs external I/O, calls an LLM, or makes non-deterministic/heuristic decisions ships a replayable append-only JSONL audit log in the same change. Use when adding or reviewing such a feature (the review question "which log answers why, and can we replay it offline?"), when designing a new audit record schema, when a recurring failure needs corpus-driven repair (replay harness, positive/negative ground truth, re...Votes: 0GitHub stars: 2
- Shadow Mode ValidationDesign pattern for shadow-mode validation — running a candidate decision mechanism (typically an LLM judgment) in observe-only parallel with the live path, recording what it WOULD have decided per event, and letting the accumulated record decide enforcement. Use when an unvalidated stochastic mechanism is about to replace or filter a live behavior (a one-way door for output quality), when a selector/classifier/gate has no published reliability evidence for the model class in play, or when des...Votes: 0GitHub stars: 2
- Codex ReviewCross-model second opinion from the OpenAI Codex CLI (a different model family), read-only, in two seams — (1) code review of the current diff; (2) plan-stage premise challenge of a design packet (refute / missing / alternative, never a design). Opt-in only — use ONLY when the user explicitly asks with "codex review", "cross-model review", "second opinion on this diff", "別モデルでレビュー", "プランを Codex に反証させて", "前提を別モデルで叩いて", or invokes /codex-review or /codex-review --plan <file>. NOT a default step...Votes: 0GitHub stars: 2
- HelperHelp with a request.Votes: 0GitHub stars: 2
- RouterRoute a request to the right helper.\n- hsfx:helper: when the request needs the helper\n- note: keep the answer shortVotes: 0GitHub stars: 2
- Adr WriterRecord a design decision as an ADR in the repo's docs/adr/, or decide whether it clears the filing bar or belongs in a note or the commit body. Use when recording a decision (ADR にして), or when context-sync finds a buried decision.Votes: 0GitHub stars: 2
- Agent StocktakeAudit the ~/.claude/agents definitions and give each a verdict.Votes: 0GitHub stars: 2
- Author Calibrated EvalPolish LLM-written prose with the author's own reading as ground truth.Votes: 0GitHub stars: 2
- Context SyncAudit a repo's context documents (CLAUDE.md, ADRs, README, graph.jsonld) for role overlap and staleness against the code, and fix them; README findings are only flagged. Use when the docs may have drifted from the code or duplicate each other. For README content, use readme-writer.Votes: 0GitHub stars: 2
- Generation AuditAudit the harness when a new Claude model takes a role (judge, build or everyday).Votes: 0GitHub stars: 2
- Harness BoundaryDesign lens: which layer a harness mechanism belongs in, and whether it outlives the next model or runtime.Votes: 0GitHub stars: 2
- Jev Judgment DesignDesign how a closed LLM judgment (relevance, novelty, evidence strength) moves to TypeSafe Jev. Use when replacing an LLM judgment with Jev, or when a Jev judgment lets everything through. For Jev's API and primitives, use typesafe:typesafe-ai first.Votes: 0GitHub stars: 2
- Learn EvalExtract a reusable pattern from this session and route it to where something reads it, or drop it. Use when saving what was learned in this session (今回の学びを残して).Votes: 0GitHub stars: 2
- Llm As JudgeDesign pattern for LLM-as-judge evaluators: binary checks as evidence, one named verdict.Votes: 0GitHub stars: 2
- Measurement DisciplineMeasurement discipline for claims that rest on data. Use when deciding on an experiment result, setting a threshold (閾値) or guard, choosing an observation period (観察期間) or when to open a gate, comparing a candidate with production (本番比較), or when an automated gate rejects outputs you trust.Votes: 0GitHub stars: 2
- Repo Asset StocktakeAudit a repo's non-code assets (tool configs, CI workflows, runbooks, docs) for ones whose consumer is gone. Use when configs, workflows or runbooks may be dead.Votes: 0GitHub stars: 2
- Review To LintMove the mechanical items of a reviewer's checklist or past findings into a deterministic script.Votes: 0GitHub stars: 2
- Rules DistillDistill principles from the installed skills into the always-loaded rules layer.Votes: 0GitHub stars: 2
- Rules StocktakeAudit ~/.claude/rules for residency cost, staleness, redundancy and substrate absorption, and give each rule a verdict. Use when the rules may have bloated.Votes: 0GitHub stars: 2
- Search FirstSearch the live web, registries and primary sources at decision time and report what exists outside. Use when choosing a library or tool, looking for a prior implementation, checking whether a paper or post applies here, reading what an official spec does now, surveying current practice, or checking a claim against its source.Votes: 0GitHub stars: 2
- Skill ComplyMeasure whether a skill, rule or agent definition is actually followed. Use when a rule's or skill's real adherence is in question, or right after adding one.Votes: 0GitHub stars: 2
- Skill CreatorDraft gate for writing or substantially revising a skill or agent definition. Use when making or revising a skill (skill 作って), writing an agent definition, or when learn-eval promotes a pattern.Votes: 0GitHub stars: 2
- Skill HealthScan the skill library for dangling references and build a per-skill health table.Votes: 0GitHub stars: 2
- Skill StocktakeAudit the installed skills and decide which to keep, improve, merge or retire. Use when taking stock of the skill library (skill の棚卸し). For writing or revising one skill, use skill-creator.Votes: 0GitHub stars: 2
- Verify BootstrapSet up a repo's machine gates (format, lint, type check, security, dependency audit, tests), or audit whether existing gates have gone stale. Use when a repo has no automated gate (lint を入れて), or its tools may be outdated.Votes: 0GitHub stars: 2
- Agent Adoption TriageFive-question triage for routing AI work to the right architecture — script, algorithmic search, LLM workflow, or autonomous agentic loop — before any agent is built. Use when someone says "let's build an agent for X", when deciding whether a use case actually needs an autonomous agent, when reviewing an agent deployment proposal or an existing agent for accountability risk, or when choosing between a pre-defined LLM workflow and a runtime-directed loop. Covers the Business AI Quadrants, the ...Votes: 0GitHub stars: 2