agentleFS
Sign inStart free
ramkesh430ramkesh430/facebook-agent-skill/skills/fb-human/SKILL.md

fb-human

Skill1 starsChanged 16 days agoMITView on GitHub

The humanizer. Strip the AI fingerprint from any Facebook draft - invisible characters, em dashes, curly quotes, stock words, Facebook engagement-bait phrases - then score what is left on a five-check panel. Use before anything in this pack is shown to the user, and whenever they say "make this sound human", "this sounds like AI", "de-slop this", or paste a draft to clean.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every other skill in this pack runs its draft through here before the user sees it.

    line 13

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

fb-human

Every other skill in this pack runs its draft through here before the user sees it. Two tools, both real:

python3 humanize.py draft.txt --report       # clean it, list every change
python3 humanize.py draft.txt -o clean.txt   # write the clean version
python3 detect.py draft.txt                  # five-check score
python3 detect.py draft.txt clean.txt        # prove the delta

What humanize.py fixes automatically

  1. Invisible characters. Zero-width spaces and joiners, word joiners, BOMs, soft hyphens, bidi controls, Unicode tag characters, odd-width spaces.
  2. Typography. Em dash -> comma, en dash -> hyphen, curly -> straight quotes, ellipsis -> three dots, bullets -> hyphens.
  3. The lexicon in slop.json: about 190 stock words and phrases with plain replacements, longest match first, case preserved. The Facebook block catches what's specific here: "like and share", "tag a friend who needs this", "type amen", "1 like = 1 prayer", "copy and paste this", "we are pleased to inform", "our valued customers", "check out our page".

slop.json is meant to be edited. Add the words the user hates.

What gets flagged, not fixed

Rewriting a sentence's shape needs judgement, so these come back to you: "It's not just X, it's Y", "not only... but also", staged one-line reveals, rule-of-three triads, fire/rocket/sparkle emoji, hashtag walls, Facebook engagement bait, "stop scrolling", video preambles, emoji bullet lists, three shouted words in a row, and press-release openers ("We are pleased to announce").

Rewrite every flag by hand in the user's voice, then re-run detect.py.

The five checks

checkwhat it measures
BURSTINESSvariation in sentence length; people vary, models don't
SPECIFICITYnumbers, names and places per 100 words
SLOP DENSITYstock terms per 100 words
FINGERPRINTinvisible characters, em dashes, curly quotes
VOICEstructural tells; under 50 caps the total at 60

65+ is PASS. Show the user the score with the draft.

Honest limits

These are local heuristics modelled on the signals public detectors look at. They are not GPTZero, Originality, Copyleaks or Turnitin, they don't call them, and they can't promise those verdicts. Nobody can honestly sell "undetectable". The invisible-character pass removes a real, checkable fingerprint; it is not a claim about defeating cryptographic watermarking.

The real goal isn't passing a detector. It's that the user's customers read the post and hear the user.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free