Claude Skill

episode

Split one source post into tier-sized drafts (teaser, main social post, longread, technical dev-log) with cross-links downward and a canonical link back to the repository. Draft-first: publishes nothing itself. Triggers: "/episode", "/adapt", "adapt this post", "make an episode".

LLM Mart · 0 points · 2 views 0 listing impressions 0 install-command copies
Download tonydzi-second-brain-starter-kit-skills_episode-7bc2efa.zip · 3 KB

Install

skills CLI npx skills add https://github.com/tonydzi/second-brain-starter-kit/tree/master/skills/episode
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install tonydzi-second-brain-starter-kit@llmmart
Git git clone https://github.com/tonydzi/second-brain-starter-kit.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole tonydzi/second-brain-starter-kit collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

EPISODE_ADAPTER (stage S5 of content factory v2 — the contract [[decision-content-pipeline-reality-show]] §7.2)

GOAL: ONE source post → an "episode" = one draft PER (tier × platform) from §7.2, with cross-links downward and canonical→GitHub baked in. Draft-first: we write drafts and publish NOTHING.

THE §7.2 CONTRACT (implemented 1:1):

tier      length          platforms (phase 1)                    language   voice/model
teaser    240-370 (HARD)  RU channel · X (EN)                    RU TG/EN X  authorial · Opus
medium    800-2500        Facebook (RU)                          RU          authorial diary · Opus · LINK IN THE 1ST COMMENT (not in the body) + a question
longread  4000+ narrative RU channel·VC.ru·Habr(RU) · GitHub(EN) RU and EN  authorial reality-show · Opus
dev-log   a dry log       GitHub (EN, canonical)                 EN          technical, impersonal · Sonnet scaffold + Opus for cohesion
  • The owner's voice (authorial) = Opus ONLY. The dev-log is not authorial — Sonnet assembles it, Opus stitches it together.
  • canonical = the GitHub markdown on every platform (a single source of truth for AI). Repo: github.com/tonydzi/clawrush.
  • NO copy-paste — every file is a NATIVE rewrite for its platform.
  • Cross-links downward: teaser→longread; medium→longread (in the 1st comment); longread ←teaser →dev-log; dev-log→longread + an anchor to our Deep Research.
  • THE PLATFORM DECIDES THE LANGUAGE (there is no bilingual pair on every tier).
  • The S3 narrative ([[style-reality-show]]) by tier: teaser=cold open+cliffhanger; medium=a full episode arc; longread=an extended episode+recap; dev-log=raw material, no drama.
  • The cofounder CTA (medium+longread, the owner's rule, memory [[cofounder-cta-public-contact]]): WhatsApp +1 341 222 9178 = an authorized PUBLIC contact (a privacy carve-out), never strip it.
  • VC.ru + Habr (the RU geo longread) = phase 1 since 2026-07-04 (the style files _STYLE-vcru.md/_STYLE-habr.md are written and live in content-factory\; the writer loads them by the pointer in the placeholder). PHASE 2 (deferred): only Reddit (the EN geo dev-log) — scaffolded with the --with-phase2 flag.

PATHS:

  • ⭐ FACTS AND CONTINUITY (v2, 2026-07-10) = THE SINGLE CANON $OBSIDIAN_VAULT/04-Projects\show-canon\: before assembling an episode, read the anchor beat (beats\) plus its arcs/loops — "previously on the show", the cliffhanger and the season question come FROM THERE (not from memory, not from SHOW-STATE, which is frozen). Respect the beat's reveal axis: don't burn live_hold/spoiler_until. Pick the next move with /reality-show next. After publishing — write the consequences back into the beat + run canon_render.py.
  • Source material: $IMPORTS_ROOT/content-factory/triage/posts.md (📝) — or a topic the owner names.
  • The scaffolder (deterministic, 0 tokens): $IMPORTS_ROOT/content-factory/episode_adapter.py.
  • Bundles: episodes\<slug>\ (8 .md files in phase 1 / 9 with --with-phase2 + meta.json; every §7.2 field is baked into the placeholders).
  • Style: [[style-reality-show]] (08-Templates\style-reality-show.md) + 04-Projects\personal\facebook-diary-auto\_STYLE.md + [[style-Mei-influence]] + the golden corpus up to 2023.

STEPS:

  1. PICK the topic: if the owner named one, use it; otherwise show the 📝 items from posts.md and take the strongest thread. Run list before new.
  2. SCAFFOLD: python episode_adapter.py new --slug <kebab> --title "<title>" --source "hub:<id>" (+--with-phase2 for VC.ru/Habr/Reddit) → a bundle of empty drafts + meta.json.
  3. WRITE (the model is named in each placeholder): teaser-ru/en, medium-fb (the body WITHOUT the link + a "1ST COMMENT" block), longread-ru/en, devlog. Replace the stubs, leave the hint comments in place. Privacy is strict (the only carve-out is the cofounder's WhatsApp).
  4. CHECK (the visibility layer): python episode_adapter.py check --slug <slug> — lengths (a teaser outside 240-370 = HARD FAIL; medium/longread = WARN), CTA present, empty files. Only FAIL=0 → step 5 (a WARN on an unfinished longread is acceptable until it is written out).
  5. REVIEW (draft-first): set-status --slug <slug> --status review; into Telegram Saved (chat_id <YOUR_CHAT_ID>, account "work_acct_a", no parse_mode): "🎬 Episode '
Files (second-brain-starter-kit)
  • SKILL.md 6.8 KB
    ---
    name: episode
    description: >-
      Split one source post into tier-sized drafts (teaser, main social post, longread, technical
      dev-log) with cross-links downward and a canonical link back to the repository. Draft-first:
      publishes nothing itself. Triggers: "/episode", "/adapt", "adapt this post", "make an
      episode".
    license: MIT
    ---
    
    EPISODE_ADAPTER (stage S5 of content factory v2 — the contract [[decision-content-pipeline-reality-show]] §7.2)
    
    GOAL: ONE source post → an "episode" = one draft PER (tier × platform) from §7.2, with cross-links downward and canonical→GitHub baked in. Draft-first: we write drafts and publish NOTHING.
    
    THE §7.2 CONTRACT (implemented 1:1):
    ```
    tier      length          platforms (phase 1)                    language   voice/model
    teaser    240-370 (HARD)  RU channel · X (EN)                    RU TG/EN X  authorial · Opus
    medium    800-2500        Facebook (RU)                          RU          authorial diary · Opus · LINK IN THE 1ST COMMENT (not in the body) + a question
    longread  4000+ narrative RU channel·VC.ru·Habr(RU) · GitHub(EN) RU and EN  authorial reality-show · Opus
    dev-log   a dry log       GitHub (EN, canonical)                 EN          technical, impersonal · Sonnet scaffold + Opus for cohesion
    ```
    - The owner's voice (`authorial`) = **Opus ONLY**. The dev-log is not authorial — Sonnet assembles it, Opus stitches it together.
    - **canonical = the GitHub markdown** on every platform (a single source of truth for AI). Repo: `github.com/tonydzi/clawrush`.
    - **NO copy-paste** — every file is a NATIVE rewrite for its platform.
    - Cross-links downward: teaser→longread; medium→longread (in the 1st comment); longread ←teaser →dev-log; dev-log→longread + an anchor to our Deep Research.
    - THE PLATFORM DECIDES THE LANGUAGE (there is no bilingual pair on every tier).
    - The S3 narrative ([[style-reality-show]]) by tier: teaser=cold open+cliffhanger; medium=a full episode arc; longread=an extended episode+recap; dev-log=raw material, no drama.
    - The cofounder CTA (medium+longread, the owner's rule, memory [[cofounder-cta-public-contact]]): WhatsApp **+1 341 222 9178** = an authorized PUBLIC contact (a privacy carve-out), never strip it.
    - **VC.ru + Habr (the RU geo longread)** = phase 1 since 2026-07-04 (the style files `_STYLE-vcru.md`/`_STYLE-habr.md` are written and live in `content-factory\`; the writer loads them by the pointer in the placeholder). **PHASE 2** (deferred): only `Reddit` (the EN geo dev-log) — scaffolded with the `--with-phase2` flag.
    
    PATHS:
    - ⭐ FACTS AND CONTINUITY (v2, 2026-07-10) = THE SINGLE CANON `$OBSIDIAN_VAULT/04-Projects\show-canon\`: before assembling an episode, read the anchor beat (`beats\`) plus its arcs/loops — "previously on the show", the cliffhanger and the season question come FROM THERE (not from memory, not from SHOW-STATE, which is frozen). Respect the beat's `reveal` axis: don't burn live_hold/spoiler_until. Pick the next move with `/reality-show next`. After publishing — write the consequences back into the beat + run `canon_render.py`.
    - Source material: `$IMPORTS_ROOT/content-factory/triage/posts.md` (📝) — or a topic the owner names.
    - The scaffolder (deterministic, 0 tokens): `$IMPORTS_ROOT/content-factory/episode_adapter.py`.
    - Bundles: `episodes\<slug>\` (8 .md files in phase 1 / 9 with `--with-phase2` + meta.json; every §7.2 field is baked into the placeholders).
    - Style: [[style-reality-show]] (`08-Templates\style-reality-show.md`) + `04-Projects\personal\facebook-diary-auto\_STYLE.md` + [[style-Mei-influence]] + the golden corpus up to 2023.
    
    STEPS:
    1. PICK the topic: if the owner named one, use it; otherwise show the 📝 items from `posts.md` and take the strongest thread. Run `list` before `new`.
    2. SCAFFOLD: `python episode_adapter.py new --slug <kebab> --title "<title>" --source "hub:<id>"` (+`--with-phase2` for VC.ru/Habr/Reddit) → a bundle of empty drafts + meta.json.
    3. WRITE (the model is named in each placeholder): teaser-ru/en, medium-fb (the body WITHOUT the link + a "1ST COMMENT" block), longread-ru/en, devlog. Replace the stubs, leave the hint comments in place. Privacy is strict (the only carve-out is the cofounder's WhatsApp).
    4. CHECK (the visibility layer): `python episode_adapter.py check --slug <slug>` — lengths (a teaser outside 240-370 = HARD FAIL; medium/longread = WARN), CTA present, empty files. Only FAIL=0 → step 5 (a WARN on an unfinished longread is acceptable until it is written out).
    5. REVIEW (draft-first): `set-status --slug <slug> --status review`; into Telegram Saved (chat_id `<YOUR_CHAT_ID>`, account "work_acct_a", no parse_mode): "🎬 Episode '<title>' — the per-tier drafts are ready (check ✅). episodes\<slug>\. Say 'publish' and I'll lay them out." NOTHING goes outward.
    6. EXPORT TO PUBLISHING — only on the owner's explicit "publish"/"+" (Tier-2): `python episode_adapter.py export --slug <slug>` → `drafts\episode-<slug>.md` (format `type: content-factory-draft` + the registers `## -> TG/X/FB`). From there the EXISTING path: `content_approve.py --serve` → `content_publish.py` (→ Saved; the real channels are Phase 2b/Tier-2). GitHub/Reddit/VC.ru/Habr are published MANUALLY (gh / natively), not through content_publish. We do NOT breed a second publisher.
    
    LIMITS:
    - Draft-first OUTWARD: channels/X/GitHub are untouched without an explicit "publish".
    - MODEL: the authorial voice = Opus ONLY; dev-log = Sonnet+Opus; scaffold/check/export = deterministic, 0 tokens.
    - One post = one episode. Windows cp1252: the script forces UTF-8 stdout itself.
    - Don't invent facts — write from the real source material.
    - Old bundles (4-file / interim) are still read by `list/show/check/export` through meta (backward compatibility).
    
    <!--kit-footer-->
    
    ---
    
    **Like this skill?** It is part of [second-brain-starter-kit](https://github.com/tonydzi/second-brain-starter-kit): the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with `npx skills add tonydzi/second-brain-starter-kit`. Everything is open source and free, so take what you need.
    
    Flagships worth a look on their own: [secondop-panel](https://github.com/tonydzi/secondop-panel) (a second opinion from a panel of external models), [claude-memory-tidy](https://github.com/tonydzi/claude-memory-tidy) (stop your agent's memory from rotting), [telegram-mcp-kit](https://github.com/tonydzi/telegram-mcp-kit) (your own Telegram over MCP in about 15 minutes).
    
    Author: **Anton Dziatkovskii**, Palo Alto AI Research Lab. Telegram [@tonydzi](https://t.me/tonydzi) - WhatsApp [+1 341 222 9178](https://wa.me/13412229178) - X [@Tony_Stef_](https://x.com/Tony_Stef_)
    
    **Engineers: want to test-drive this setup?** Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.