Track and update action item status from natural language updates like 'Seanのメールには返信しておいた' or 'Lu対応予定'. Integrates with daily-comms-ops workflow. Use when managing action items across communication channels.
skill-idea-miner
Mine Claude Code session logs for skill idea candidates. Use when running the skill generation pipeline to extract, score, and backlog new skill ideas from recent coding sessions.
含まれるファイル(7)
- SKILL.md2.9 KB
- references/idea_extraction_rubric.md6.3 KB
- scripts/mine_session_logs.py24.4 KB
- scripts/score_ideas.py15.1 KB
- scripts/tests/conftest.py223 B
- scripts/tests/test_mine_session_logs.py30.8 KB
- scripts/tests/test_score_ideas.py9.9 KB
SKILL.md(原文)
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Skill Idea Miner
Automatically extract skill idea candidates from Claude Code session logs, score them for novelty, feasibility, and work utility, and maintain a prioritized backlog for downstream skill generation.
When to Use
- Weekly automated pipeline run for skill idea generation
- Manual backlog refresh:
python3 scripts/mine_session_logs.py - Dry-run to preview candidates without LLM scoring
Workflow
Stage 1: Session Log Mining
- Enumerate session logs from project directories in
~/.claude/projects/ - Filter to past 7 days by file mtime, confirm with
timestampfield - Extract user messages (
type: "user",userType: "external") - Extract tool usage patterns from assistant messages
- Run deterministic signal detection:
- Skill usage frequency (
skills/*/path references) - Error patterns (non-zero exit codes,
is_errorflags, exception keywords) - Repetitive tool sequences (3+ tools repeated 3+ times)
- Automation request keywords (English and Japanese)
- Unresolved requests (5+ minute gap after user message)
- Skill usage frequency (
- Invoke Claude CLI headless for idea abstraction
- Output
raw_candidates.yaml
Stage 2: Scoring and Deduplication
- Load existing skills from
skills/*/SKILL.mdfrontmatter - Deduplicate via Jaccard similarity (threshold > 0.5) against:
- Existing skill names and descriptions
- Existing backlog ideas
- Score non-duplicate candidates with Claude CLI:
- Novelty (0-100): differentiation from existing skills
- Feasibility (0-100): technical implementability
- Work Utility (0-100): practical value for business professionals
- Composite = 0.3 * Novelty + 0.3 * Feasibility + 0.4 * Work Utility
- Merge scored candidates into
logs/.skill_generation_backlog.yaml
Output Format
raw_candidates.yaml
generated_at_utc: "2026-03-08T06:00:00Z"
lookback_days: 7
sessions_analyzed: 12
candidates:
- id: "raw_20260308_001"
title: "Project Charter Generator"
scanned_projects: ["PycharmProjects-claude-skills-library"]
description: "Generate project charters from requirements."
category: "project-management"
Backlog (logs/.skill_generation_backlog.yaml)
updated_at_utc: "2026-03-08T06:15:00Z"
ideas:
- id: "idea_20260308_001"
title: "Project Charter Generator"
description: "Skill that generates project charters..."
category: "project-management"
scores: {novelty: 75, feasibility: 60, work_utility: 80, composite: 72.5}
status: "pending"
Resources
references/idea_extraction_rubric.md-- Signal detection criteria and scoring rubricscripts/mine_session_logs.py-- Session log parserscripts/score_ideas.py-- Scorer and deduplicator
レビュー
まだレビューはありません。使ってみた感想をお寄せください。
同じリポジトリのスキル
概要と使いどころ
AI活用提案の専門コンサルタントスキル。LLM・AIエージェントの活用事例知識を基に、業界別、部門別、シーン別に最適なAI導入提案を行います。システム化、業務改善、効率化を得意とし、具体的な活用事例、期待効果、導入ステップを含む実践的な提案を提供します。Use when proposing AI/LLM adoption strategies for specific industries, departments, or business scenarios.
Generate AI-powered BPO service proposals for Japanese companies in the US market. Use when creating outsourcing proposals with AI service menus, ROI estimation, implementation roadmaps, and bilingual (JA/EN) proposal documents for 在米日系企業向けAI実装.
日本語の概要は準備中です。原文の説明を表示しています。
AI(LLM)が生成した日本語テキストの「AI臭」を検出・診断し、人間らしい文章にリライトするスキル。 Use when: 「AIっぽい文章を直して」「人間らしくリライトして」「AI臭を消して」「この文章をもっと自然にして」 「テキストのAI感を減らして」「機械っぽさを取りたい」 "make this sound more human", "remove AI tone", "humanize this text", "detect if this is AI-generated". 6つのAI特有パターン(視覚的マーカー、単調なリズム、マニュアル的構成、非コミット姿勢、抽象語の濫用、 定型メタファー)を正規表現ベースで検出し、0-100のAI臭スコアを算出。3つの人間化技法 (バランスを崩す・客観を崩す・論理を崩す)でリライトを実行する。 Note: 検出スクリプトは日本語テキスト専用。英語テキストの場合はClaude自身がreferences/を参照して分析・リライトする。
Generate audit-ready internal control design documents from As-Is business process inventories. Produces control IDs, assertion mappings, procedures, SoD analysis, KPIs, materiality thresholds, and implementation roadmaps. Use when building internal controls for new audit engagements, SOX/J-SOX compliance, or process improvement initiatives.
日本語の概要は準備中です。原文の説明を表示しています。
Review audit-related documents (control design documents, bottleneck analyses, requirements definitions, etc.) for quality, scoring them 0-100 with a severity-rated findings list. Use when reviewing audit documents, checking control design quality, or verifying cross-document consistency. Supports documents governed by US GAAP, IFRS, or J-GAAP.
日本語の概要は準備中です。原文の説明を表示しています。