本文へ移動
cccskills
無料GitHub で公開

series-planner

Plan multi-part content series: structure, cross-linking, cadence.

インストール方法を見る

含まれるファイル(5)

  • SKILL.md4.0 KB
  • references/cadence-guidelines.md6.8 KB
  • references/cross-linking.md7.2 KB
  • references/output-format.md3.8 KB
  • references/series-types.md7.6 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。

Series Planner Skill

Plan multi-part content series: assess viability, select structure, produce plan with cross-linking and cadence. Three phases with strict gates.

Deep References

SignalLoadWhy
Publishing frequency, delay handlingreferences/cadence-guidelines.mdSchedule selection criteria
Navigation links, Hugo implementationreferences/cross-linking.mdPrev/next nav, shortcodes, landing pages
Plan output formatreferences/output-format.mdComplete plan template
Series types, selection matrixreferences/series-types.mdThree type templates with examples

Usage

/series [topic]
/series --type=progressive [topic]     # Force type
/series --parts=5 [topic]              # Target count
/series --with-landing [topic]         # Include landing page
/series --minimal [topic]              # Titles and scope only

Phase 1: ASSESS

Determine viability and natural divisions.

  1. Analyze topic: scope (narrow/medium/broad), natural divisions, audience progression.
  2. Check viability: minimum 3 distinct subtopics, each stands alone, logical progression exists, no filler.
  3. Detect series type:
SignalType
learn, master, deep diveProgressive Depth
build, create, projectChronological Build
why we chose, migration, debuggingProblem Exploration

Gate: 3+ natural divisions identified. If fewer, recommend single post or scope adjustment.

Phase 2: DECIDE

Select type, part count (strict 3-7 bounds), and structure.

  1. Select type and justify. Define part count and word estimate.
  2. Draft breakdown per part: title, scope (1 sentence), standalone value, adjacent links.
  3. Validate standalone value for every part:
    • Reader learns something complete and actionable from this part alone
    • Working code/config/output possible without other parts
    • No critical info deferred to other parts
    • Search landing on this part alone yields value

Red flags (reject): "read Part 1 first" dependency, mid-implementation endings, concepts only in earlier parts.

  1. Select cadence (see references/cadence-guidelines.md). Default: weekly.

Gate: All parts pass standalone test. Part count 3-7. Type justified.

Phase 3: GENERATE

Produce the complete plan.

  1. Build plan: series header, per-part breakdown, cross-linking (see references/cross-linking.md), publication dates, Hugo frontmatter.
  2. Final validation checklist:
    • Every part has standalone value
    • Word counts 800-1500/part, within 20% variance
    • Cross-linking complete (prev/next for all)
    • No cliff-hangers, no filler
    • Part count within bounds
  3. Output using references/output-format.md format.

Gate: All checks pass. Plan complete.

Series Types

TypeStructureReader flexibility
Progressive DepthShallow to deep masteryBeginners stop at Part 1; advanced skip ahead
Chronological BuildStep-by-step creationEach part produces working output
Problem ExplorationProblem to solution journeyEven failed approaches are instructive

Error Handling

ErrorCauseSolution
Topic too narrowFewer than 3 natural divisionsRecommend single post or scope expansion
Topic too broadWould need 8+ partsSplit into multiple series; narrow scope
No logical progressionParts are loosely related, not buildingConsider standalone posts instead of series
Standalone value missingParts depend on each otherMerge dependent parts or add context inline

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

accounts

無料

Manage multiple Claude Code accounts: add, list, check, launch, and install shell aliases for 10+ isolated CLAUDE_CONFIG_DIR profiles.

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。

notque/vexjoy-agent4412026年10月12日 更新

Improve architecture across modules by deepening interfaces.

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。

notque/vexjoy-agent4412026年10月12日 更新

Assessment: read-only inspection, codebase overview, value analysis, health checks, ADR consultation, decision analysis, multi-perspective critique.

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。

notque/vexjoy-agent4412026年10月12日 更新

Background memory consolidation — overnight review, merge, and injection payload for memory files.

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。

notque/vexjoy-agent4412026年10月12日 更新

Jev-driven browser automation: Jev picks operations, programs execute, a text model writes field values only when Jev cannot pick one from the goal.

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。

notque/vexjoy-agent4412026年10月12日 更新

Write, compose, integrate, and improve programs that call Jev, TypeSafe's System One judgment model.

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。

notque/vexjoy-agent4412026年10月12日 更新

notque のスキルをすべて見る

このスキルの問題を報告する