All skills
simota avatar

/tome

@35ffd55
by shingo imotasimota/agent-skills85 stars
15

Converting technical knowledge into durable learning documents and publishable articles. Use for diff-based teaching, decision records, onboarding, note/Zenn/Qiita/dev.to posts, article series, retrospectives, and cross-platform repurposing.

Use this Skill: https://skilld.dev/gh/simota/agent-skills/tome

This session only. Nothing lands on disk.

referencearticle-series-management.md

≈3.2k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Series Management

Purpose: Multi-episode tech blog series require different discipline than one-shot articles — index coherence, cross-links, cadence, tonal continuity. Treat series as a product, not a playlist. Read when: FRAME phase (new series kickoff or integrating an episode into existing series) and PUBLISH phase (updating index + cross-links).

Why Series Need Their Own Discipline

A standalone article has one job: stand alone. A series episode has three:

  1. Stand alone enough that first-time readers get value without reading prior episodes.
  2. Reward readers who have been following — callbacks, terminology continuity, narrative threads.
  3. Serve the series arc — advance the thesis, don't just exist as another post tagged "series X".

Skip any of these and the series breaks. #1 broken = new readers bounce. #2 broken = loyal readers feel ignored. #3 broken = the series becomes a bag of loosely-related posts with the same prefix.


Index Article Design

The index article (often named #00 Overview, Part 0: Introduction, or similar) is the anchor readers return to. It's also the single most-updated file in the series.

What the index article contains

# [Series Title]

**[One-line thesis of the series]**

[200-400 chars: what this series argues, who it's for, what reader will take away]

## Episode List

- [x] #00 Overview — this article
- [x] #01 [Title] — [one-sentence teaser]
- [x] #02 [Title] — [one-sentence teaser]
- [ ] #03 [Title] — [planned teaser, link added on publish]
- [ ] #04 [Title] — (planned)

## How to Read This Series

[Reading order: chronological / any order / must-read first 3 / etc.]

## Release Cadence

[Weekly Monday 10:00 JST / Biweekly / As-ready — state explicitly]

## About

[Author context, why this series exists]

Update discipline

Every new episode must, in the same pass:

  1. Add the episode to the index's Episode List (change [ ] to [x], add link).
  2. Update the index's "How to read" if the new episode changes the reading order (e.g., "must-read first 3" changes to "first 4").
  3. Consider whether the series thesis paragraph needs refinement — a mid-series episode sometimes clarifies the thesis, and the index should reflect.
  4. Commit both files together (feat(series): publish #09 Forge and update index).

Anti-pattern: Publishing the episode and "updating the index later". Later never happens. Readers who land on the index see an outdated list and lose trust.


Cross-Reference Strategy

Top-of-episode navigation

Every non-index episode opens with a nav strip:

> 連載「[Series Title]」第N回。前回: [#N-1 Title]。
> 目次: [#00 Overview]

For English (dev.to):

> Part N of the "[Series Title]" series. Previous: [#N-1 Title].
> Index: [#00 Overview]

Bottom-of-episode navigation

Every non-index episode closes (before CTA) with:

---

- 前回: [#N-1 Title]
- 次回 (予定): [#N+1 Working Title] — [one-line teaser]
- 目次に戻る: [#00 Overview]

次回 (予定) = tentative. If the series might skip around, say so. If a future episode's title is locked, link it directly.

In-body back-references

When episode #07 refers to a concept established in #03, link explicitly:

([#03 Concept](link) で触れた「〇〇」を、ここでは実装に落とす。)

Readers who skipped #03 have a way back. Readers who remember #03 get a little dopamine hit of continuity. Both wins.


Release Cadence

State the cadence explicitly in the index article so readers know what to expect.

Cadence Pros Cons Fit
Weekly (fixed day/time) Reader habit, RSS compatibility, predictable Pressure, may sacrifice quality for schedule Established series, author with reliable time
Biweekly Breathing room for research-heavy episodes Readers may lose context between episodes Deep-dive technical series
Burst (2-3 in a week, gap, burst) Natural for topic clusters, momentum Unpredictable, RSS readers get clumps Narrative arcs, project retrospectives
As-ready Quality over schedule Series drifts, readers forget it exists Experimental series, side projects

Commitment lever: If you're writing publicly, a stated weekly cadence forces discipline. The cost is burn-out; the benefit is compounding readership.

Recovery lever: When a stated cadence slips (missed a week), address it in the next episode opening: "先週は休載した。理由は〜。今週の #07 から再開する。" Readers respect honesty more than silent gaps.


Naming Convention

Pick one and commit for the entire series:

Convention Example When it fits
#NN Title #00 Overview, #01 Concept JP / note convention, short and scannable
Part N: Title Part 0: Introduction, Part 1: Setup English / dev.to convention
Episode NN — Title Episode 01 — Concept Longer form, works for narrative-heavy series
[Series]: Title Agent Skills: Forge When episode order doesn't matter

Hard rule: Don't change the convention mid-series. If you started with #00, you're committed to #NN for the arc. Changing conventions mid-series (e.g., switching to Part N at episode #05) breaks search, breaks bookmarks, breaks the index.

Zero-indexing vs one-indexing: #00 Overview + #01..#N treats the overview as a "part zero" meta-episode. #01 Overview + #02..#N treats overview as the first real episode. Both are valid; pick one and be consistent.


Tonal Continuity (Series Bible)

Multi-episode series need a lightweight series bible to maintain consistency. Store it in .agents/PROJECT.md or a dedicated .agents/tome-article-series-{name}.md:

## Series Bible: [Series Title]

### Voice
- First person / third person: [choice]
- Formality: [です・ます / だ・である / casual]
- Reader address: [あなた / 読者の皆さん / none]

### Recurring terminology
- エージェント (not AI / bot / 助手)
- スキル (not 能力 / capability)
- [Project-specific terms with definitions]

### Cast of characters (if any)
- [Character name]: [role, first appearance]
- [Character name]: [role, first appearance]

### Recurring metaphors
- [Metaphor]: [when introduced, what it represents]

### Phrases to avoid (beyond universal anti-patterns)
- [Series-specific throat-clearing to cut]

### Episode cadence
- Target: [weekly Monday 10:00 JST]
- Grace period: [+3 days before officially "slipped"]

### Thesis
[The series' core argument, 1-3 sentences — this is what every episode should serve]

Before drafting a new episode:

  1. Re-read the series bible.
  2. Re-read the previous episode's last paragraph (continuity check — does the new episode open naturally from where the last one ended?).
  3. Check terminology drift — did the last episode start using a word the bible doesn't sanction?

Finale vs Open-ended

Decide at series kickoff:

Finale-ending series

  • Stated episode count from the start (e.g., "全10回の連載").
  • Builds toward a concluding thesis, synthesis, or artifact (a book, a framework, a project).
  • Final episode explicitly wraps: "#10 結論" or "Part 10: Conclusion".
  • Risk: Padding. If the thesis is solid at #6, don't extend to #10 for the sake of commitment.

Open-ended series

  • No stated end. Published as long as material exists and author interest holds.
  • Risks series drift — becomes a catch-all tag instead of a coherent arc.
  • Mitigation: periodic "state of the series" recap episodes (every ~10 episodes) to re-establish thesis and re-hook readers.
  • Example: "Claude Skills 図鑑" is open-ended — each episode covers one agent, there's no fixed count.

Hybrid: arc-based open-ended

  • Open-ended but organized into explicit arcs.
  • "Part 1 (Episodes #01-#10): Fundamentals" → "Part 2 (#11-#20): Advanced".
  • Each arc has a mini-finale, but the series continues.

Downstream Conversion: Series as Artifact

A completed or mature series is prime material for downstream artifacts. Plan this from #00, or at least from mid-series:

Downstream artifact How series feeds it Handoff agent
PDF zine / ebook Concatenate + polish + add cover + TOC Scribe (format conversion)
Zenn Book (paid) Restructure as chapters, polish continuity Self (reformat) + Scribe
Conference talk deck Extract thesis + 3-5 strongest episode arcs Stage (slides)
Marketing narrative / customer story Reshape as product story, not tech retrospective Saga
Internal onboarding doc Reverse-flow: public series → internal training Tome
Podcast / audio series Re-record with conversational adaptation (External tool)

Plan the anthology at #00: "This series may become a PDF zine / talk / book." The anticipation changes how you write individual episodes — you write with the eventual anthology in mind (callbacks that work in anthology context, not just feed context).


Live Example: "Agent Skills 図鑑" (user's current project)

Context: note series, #00〜#08 完成, next #09 Forge.

Observed conventions (inferred from memory context — verify in the project journal):

  • Naming: #NN AgentName
  • Cadence: as-ready (not weekly)
  • Platform: note (Japanese, long-form)
  • Open-ended: one episode per agent, series continues as new agents land

When drafting #09 Forge:

  1. Re-read #08's closing paragraph — continuity check.
  2. Re-read #00 Overview — has the agent list been updated to include Forge as "planned"? Update now.
  3. Check .agents/PROJECT.md for any series-specific terminology or tone conventions established #00-#08.
  4. Draft #09 Forge with:
    • Top nav: 前回 [#08], 目次 [#00]
    • Bottom nav: 前回 [#08], 次回 (予定) [#10 ???], 目次に戻る
  5. On publish: update #00 Overview to mark #09 as published, and (if known) add #10 as planned with working title.
  6. Commit: feat(tome): publish #09 Forge and update series index.

Series Kickoff Checklist

When starting a new series:

  • Decide platform (note / Zenn / Qiita / dev.to — platform features vary for series)
  • Decide naming convention (#NN / Part N / other) and commit
  • Decide cadence (weekly / biweekly / burst / as-ready) and state it in #00
  • Decide ending (finale count / open-ended / hybrid-arc)
  • Draft #00 Overview with thesis, episode list (all known episodes as planned), cadence, how-to-read
  • Create series bible in .agents/PROJECT.md or .agents/tome-article-series-{name}.md
  • Set up magazine (note) / series (dev.to) / topic (Zenn) for automatic grouping
  • Plan downstream anthology (PDF? book? talk? — or explicitly decide "just episodes")

Per-Episode Checklist

For every episode after #00:

  • Re-read series bible
  • Re-read previous episode's last paragraph
  • Check for terminology drift
  • Draft with top-nav and bottom-nav strips
  • Update #00 Overview's episode list (planned → published)
  • Update #00's "how to read" if reading order changed
  • Update previous episode's bottom-nav if it had 次回 (予定): TBD — fill it in now
  • Add to magazine (note) / series (dev.to)
  • Commit with feat(series): publish #NN Title and update index

Core Elements (SKILL.md excerpt)

  • Index article (e.g., #00 Overview) serves as anchor readers return to — must list all episodes with one-sentence teasers and update on every new episode.
  • Cross-links at top and bottom of each episode: 前回 → / → 次回, plus "see episode #3 for background".
  • Naming convention: #NN タイトル or Part N: Title. Pick one and stay consistent across the arc.
  • Release cadence: weekly (discipline but pressure), burst (2-3 in a week, then gap), as-ready (no commitment). State the cadence in the index article so readers know what to expect.
  • Tonal continuity: series bible (stored in .agents/PROJECT.md or journal) locks first/third-person, formality, recurring metaphors, character references across episodes.
  • Finale vs open-ended: decide at series kickoff. Open-ended needs periodic "state of the series" recap episodes.
  • Downstream conversion: a completed series is prime material for a PDF zine, paid magazine, or talk deck — plan the anthology from #00.

Source: SKILL.md on GitHub

No alerts13d4 checks · Risk SAFE
  • Gen Agent Trust Hub13d

    The 'tome' skill is a technical documentation and article authoring tool designed to convert git changes and raw source material into structured learning assets. It incorporates robust safety standards, such as a strict 'Documents only' policy and mandatory masking of sensitive information like credentials and client details. While the skill processes external data, it includes explicit mechanisms to prevent the execution of untrusted instructions, ensuring a secure environment for technical writing.

  • Socket13d

    No alerts

  • Snyk13d

    Risk: LOW · No issues

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at 35ffd55. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 3 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/tome