Skip to content

opensite-ai/opensite-skills / book-recorded-sessions

book-recorded-sessions

Verified

Plan, record, and merge recorded sessions into book chapters: solo lessons, interviews with a proxy reader, multi-guest panels, or hybrid lesson-plus-conversation episodes that also serve video and short-form clips. Covers contract-driven run-of-show, contributor releases and employer clearance, per-speaker capture and transcripts, speaker-attributed atoms, merging many sessions into a traceable chapter, and clip hand-off flags. Use when chapters are built from spoken recordings or podcasts. Not for narrating the finished audiobook or editing clips.

FileKindSize
SKILL.mdskill md4.9 KB
assets/segment_types.yamlconfig5.1 KB
references/contributor-release.mdmarkdown2.8 KB
references/merge-rules.mdmarkdown2.4 KB
references/session-formats.mdmarkdown3.8 KB
scripts/chapter_trace.pyexecutablescript8.1 KB
scripts/session_ingest.pyexecutablescript9.4 KB
scripts/session_plan.pyexecutablescript8.1 KB

references/session-formats.md

View on GitHub
# Session formats, roles, capture, and time budget

## Formats

| Format | Who | Book yield | Video and clips | Use when |
|---|---|---|---|---|
| Solo | Author to camera | Sequenced teaching; the author skips steps experts find obvious | Tutorial video, one-idea clips | Methods heavy on screen demos |
| Interview | Author plus proxy reader | Best yield of the author's own words; the questioner catches skipped steps | Tutorial or conversation | Default when there are no guests |
| Panel | Author plus practitioners | Stories, cases, disagreement; the author talks less | Strongest conversation clips | Case-driven chapters; companion episodes |
| Hybrid | Part A: lesson with a proxy reader. Part B: panel | The spine from A, cases from B | A works as a tutorial, B as a podcast episode, clips from both | Teaching books whose author also runs a show |

Roles (see `assets/segment_types.yaml`): **author** teaches and owns the argument; **proxy
reader** asks what the target reader would ask (choose someone closest to the reader, for
example an operations leader for a book aimed at non-engineers); **practitioner** brings real
cases with numbers; **producer** runs time, markers, levels, and confidentiality. People on the
show are contributors, not validators: reader evidence still comes from strangers in the
target audience.

## Time budget

Professional benchmark: ghostwriting firms that build books from author interviews record about
20-40 hours of the author talking. Scribe's standard package uses ten two-hour interviews, its
largest twenty ([Scribe](https://scribemedia.com/pricing/professional),
[Scribe Elite](https://scribemedia.com/pricing/elite)). What matters is the author's own words.
Estimate them from expected talk share (measure it with `session_ingest.py` after the first
session): a lesson or interview yields most of an hour as author talk; a three-person panel
yields roughly half or less. The coverage check (`atom_coverage.py`), not an hour count, decides
when there is enough material.

## Capture spec

- One microphone per speaker, each on its own track; WAV, 48 kHz or higher, 24-bit. Per-track
  transcripts avoid diarization errors and make attribution exact.
- Slate at the start (session id, date, chapter); 30 seconds of room tone with everyone silent.
- On the record at the start: participants confirm the recording feeds the book, audiobook,
  video, podcast, and clips under their signed release.
- The producer drops markers at each segment boundary (segment ids from the run-of-show) and
  exports `markers.csv` with `segment_id,start`.
- Back up the raw tracks before editing; never transcribe from a compressed social export.
- These recordings are source material. The audiobook is narrated separately from the locked
  text under `book-audiobook-production`.

## Designing segments that also clip well

Open each segment with one sentence that stands alone (a claim, a question, or a number), avoid
"as I said earlier", put the model on screen during the lesson, and keep real disagreement
rather than staging it. These are production habits, not measured rules; judge them with the
channel's own retention and engagement data.

## Publishing coordination

- Episodes can go out before the book; they build the audience. Keep volatile tool details out
  of the durable lesson segments.
- KDP Select, if ever used, restricts the ebook only: video, audio, and print may be distributed
  elsewhere, and up to 10% of the book may appear as a sample
  ([KDP](https://kdp.amazon.com/en_US/help/topic/GD9PMU58BV24QFZ7)). Don't publish full
  chapter text as transcripts if the ebook will be enrolled.
- Before posting a clip, clear the `segments.csv` flags. When colleagues praise the book in
  posts, the relationship is a material connection and must be clear to viewers
  ([FTC](https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking)).
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 194kUpdated Claude CodeCodexVerifiedInstall