agentsclimarketplace

Jiang transcript boundary review

Skill apresmoi/jianglens/.codex/skills/jiang-transcript-boundary-review

Agentic Organization Research Project - Jiang Lens is an independent research and reading project built from Jiang Xueqin's lectures, interviews, and writing.

Install
npx -y skills add apresmoi/jianglens --skill jiang-transcript-boundary-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 7 stars7 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use this skill when reviewing Jiang Lens transcript boundary candidates from ops/scripts/repair-transcript-boundaries.mjs and producing approve/reject decisions without rewriting the transcript or loading the full transcript into context.

SKILL.md

2.6 KB, 563 tokens by cl100k_base, as published. Nobody here has run it

Jiang Transcript Boundary Review

Use this after ops/scripts/repair-transcript-boundaries.mjs creates:

content/workflow/tasks/<source-slug>/transcript-boundary-candidates.jsonl

The goal is only to decide whether adjacent transcript segments should exchange a few words. Do not rewrite Jiang's prose. Do not edit transcript files directly.

Workflow

  1. Read candidates with shell tools, not by loading the whole transcript:
wc -l content/workflow/tasks/<source-slug>/transcript-boundary-candidates.jsonl
jq -c '{candidate_id,boundary,type,text,prev:{id,tail_text},next:{id,head_text}}' content/workflow/tasks/<source-slug>/transcript-boundary-candidates.jsonl
  1. For each candidate, decide from the provided prev.tail_text, next.head_text, and chunk texts.
  2. Write decisions to:
content/workflow/reviews/<source-slug>/transcript-boundary-decisions.json

Use this shape:

{
  "source_slug": "<source-slug>",
  "decisions": [
    {
      "candidate_id": "seg-0006->seg-0007:move-trailing-starter-to-next:b344d80e24a3",
      "decision": "approve",
      "reason": "Moves 'The' before 'simplest way...' to complete the sentence."
    }
  ]
}

Allowed decisions are approve and reject.

  1. Apply approved decisions:
node ops/scripts/repair-transcript-boundaries.mjs \
  --source content/sources/videos/<source-slug> \
  --decisions content/workflow/reviews/<source-slug>/transcript-boundary-decisions.json \
  --write
  1. Run:
node ops/scripts/compile-content.mjs
node ops/scripts/validate-content.mjs

Approval Rules

Approve when:

  • A segment ends mid-sentence and the next segment begins with the sentence completion.
  • A segment ends with a sentence starter such as The, And this, or But then, and the next segment begins with lowercase continuation text.
  • The move improves reading while preserving exact words and chronology.

Reject when:

  • The proposed text is a useful new sentence rather than a completion.
  • The move would hide an actual topic break, speaker change, or quoted reading boundary.
  • The candidate appears to be ASR noise, punctuation damage, or abbreviation damage such as .S..
  • You cannot tell from the provided two-segment window.

Prefer rejection when uncertain. This pass only fixes obvious boundary splits.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.