Index of every page on this site: llms.txt
All skills
hktitan avatar

/khe-content-engine

@945e23c
by Harshit Khemanihktitan/skills47 stars
2
Run it yourself

Builds and runs a content engine: a folder of small linked markdown nodes in your own project (a voice sampled from posts you wrote, audience segments, one node per platform you use, hooks, repurposing rules, content types, cadence), then turns one idea into a different draft for each platform and names the node behind every choice. Also reviews drafts against the engine and tunes the node that caused a wrong line. It drafts; a person posts. Use when someone says "build me a content engine", "turn this into posts for X and LinkedIn", "repurpose this", "write in my voice", "make a thread out of this", "why doesn't this sound like me", or wants one idea written for several platforms. For a launch film use `khe-launch-film`, for a product demo `khe-demo`, for stickers `khe-sticker`, for faux product UI `stageshow` (Benji Taylor). When unsure which skill owns a job, `khe-auto`.

Use this Skill: https://skilld.dev/gh/hktitan/skills/khe-content-engine

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

README.md

≈1.8k tokens on demand. Your agent reads this file only when SKILL.md points to it.

khe-content-engine

Builds a content engine in your project and turns one idea into a different draft for each platform. It drafts; you post.

npx skills add HKTITAN/skills --skill khe-content-engine

What it does

  • setup: builds the engine for one person or brand as a folder of small linked markdown nodes in their own project: an index, a voice node made of quoted samples of what they wrote and a list of what they never say, audience segments, one node per platform they use, hooks, repurposing rules, content types and cadence. It reads the posts and pages they hand over and interviews for what it cannot see, instead of inventing a voice.
  • run: takes one idea and writes one draft per platform, each a different treatment of the idea, with the node behind every choice named on the draft.
  • review: judges existing drafts against the engine and names the node to change for each fault.
  • tune: when a draft is wrong, changes the node that caused it, dates the change and runs again.
  • Marks any number or quotation the brief lacks as [confirm] instead of making one up.
  • Checks every draft for the phrases and patterns that read as generated.
  • Never publishes or schedules anything. A person reads, approves and posts.
  • Ends every reply with what it could not verify.

How it runs

khe-content-engine's flow: posts you wrote become the engine of small linked nodes; one idea runs through it into a draft per platform, with views from khe-brain-harshit or the rest of a post from khe-package; a review sends a wrong line to the node that caused it, which is tuned and rerun; a person posts

Used

/khe-content-engine run moving the due date to the top of the invoice for x, linkedin, newsletter

Three drafts came back, one treatment each, every figure traced to the owner's brief. The X draft, from examples/runs/invoice-date.md:

Median time to payment for 14 of my clients went from 38 days to 24.

They changed one line on the invoice. The due date moved to the top, written as a date.

"Net 30" is a sum. Clients do not do sums for you.

Treatment: the single claim, with its one number. Hook: the number. The LinkedIn draft tells how it was found out and what it does not prove; the newsletter gives the wording to copy. The run ends "Not verified: the figures, taken as Ines gave them from her spreadsheet; whether the illustrator approved this exact wording; how any of the three reads on a phone." content-check on it ends content-check passed.

When to use it

Use it when one person's ideas need to reach several platforms without becoming the same post several times, or when drafts written for someone keep failing to sound like them. It writes text only. For a launch film use khe-launch-film, for a product demo khe-demo, for stickers khe-sticker, and for faux product UI stageshow (Benji Taylor).

Invoke with

/khe-content-engine [setup | run <idea> for <platforms> | review <drafts> | tune <fault>]

Or ask in your own words: "build me a content engine", "turn this into posts for X and LinkedIn", "repurpose this", "write in my voice".

Before and after

Before: "turn this into posts for X and LinkedIn" returns one post and a longer copy of it, both opening the same way, with a percentage nobody measured, a client quotation nobody said and "let's dive in".

After: a run file with the brief, an X draft that states the single claim with its one figure, a LinkedIn draft that tells how it was found out and what it does not prove, each naming its treatment, hook type and nodes, then Gate: content-check passed, Posted: nothing. A person posts. and Not verified: the figures as the owner gave them.

Example prompts

  • Build me a content engine. Here are twelve of my LinkedIn posts and my newsletter archive.
  • Turn this into posts for X and LinkedIn: we moved the due date to the top of our invoices and got paid two weeks sooner.
  • Repurpose last week's newsletter issue for X. Write in my voice.
  • These drafts don't sound like me. Find out why and fix the engine, not the drafts.

Verify

node scripts/content-check.mjs content-engine --gate               # an engine folder
node scripts/content-check.mjs content-engine/runs/idea.md --gate  # a run
node scripts/content-check.mjs examples --gate                     # the gold engine and runs
node scripts/content-check.mjs --self-test                         # the gate against its own fixtures

content-check.mjs names every fault as a line. --gate ends content-check passed only when every never-intentional line is at 0, exits 1 with the counts otherwise, and exits 2 with inconclusive when it could not look.

What it proves, for an engine: every required node exists, every link between nodes resolves, no node is orphaned, the voice node has at least three sourced samples and a never-say list, every platform node has a length limit and a dated-facts line, and no template blank is left. For a run: one draft per platform asked for, each inside its platform node's limit, no two drafts with the same first line, the same treatment or more than 20% of their sentences in common, every draft names its nodes, no banned phrase or pattern, no number or quotation that is absent from the brief and unmarked, and a Not verified: last line.

What it does not prove: that a draft sounds like the person, that the hook is paid off, that two drafts differ in angle, that a figure in the brief is true, that a platform fact is still current, or that anyone agreed to be quoted. It reads text and never hears a voice. Passing is necessary, not sufficient.

--self-test also proves the lines that have no teaching pair, on broken copies of the gold engine, and checks that every claim line in references/x-algorithm.md carries a file path.

What is inside

  • SKILL.md: the one idea, the four jobs with their steps, the hard rules, the reference table and the output format.
  • references/: nodes.md (the folder, each node's template, the brief, the run file), voice-sampling.md, platforms.md (one section per platform, facts dated and graded), x-algorithm.md (what the published X ranking code states, by file), hooks.md, repurposing.md, tells.md (the banned list the gate reads), verify.md, review.md, teaching.md, sources.md, and gotchas.md and pov.md, which are yours to edit.
  • examples/: one gold engine of ten nodes for an invented bookkeeper, and four gold runs: three platforms, a thread, a client quotation with an open mark, and a skipped platform.
  • teaching/: fifteen bad/good pairs, each named for the gate line its bad half raises. They are the gate's tests.
  • scripts/: content-check.mjs (the gate). Node 20 or newer, no dependencies.
  • LICENSE: MIT. It covers this skill's own text and code.

Sources: method adapted from Ronin's public X article on building a content engine (the saved copy carried no link to the post); the X ranking reference is a reading of xai-org/x-algorithm, Apache License 2.0, at commit 660cd74; platform limits from the pages listed in references/sources.md, checked 2026-11-10. Not written or endorsed by any of them.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

1 warning4d3 checks · Risk SAFE
  • Gen Agent Trust Hub4d

    The skill is a content creation tool that builds a 'content engine' from user-supplied writing samples and drafts platform-specific posts. It includes a local Node.js script to perform static analysis and linting on generated files to ensure structural integrity and quality. No network requests, sensitive file access, or remote code execution patterns were detected.

  • Socket4d

    No alerts

  • Snyk4d

    Risk: MEDIUM · 1 issue

Signed by skilld at 945e23c. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 weeks ago
metadata
{
  "author": "Harshit Khemani",
  "version": "1.0.0"
}

README badge

README badge for hktitan/skills/khe-content-engine