LinkedIn Skills That Clean the AI Slop Before You Copy-Paste

What it does

Eleven Claude skills cover the full LinkedIn workflow: posts spun from 21 hook formulas, comments, replies, carousels, DMs, inbox triage, profile scoring, and weekly planning. Every skill ends at a copy-ready block; nothing touches LinkedIn without your manual paste. The standout is /li-human, a pair of local Python scripts that strip invisible characters, em dashes, curly quotes, and 113 stock AI phrases, then score the draft against five local heuristics before you see it.

The interesting bit

The README spends as much time explaining what the tool cannot do as what it can: it will not auto-post because that violates LinkedIn’s terms, and it will not promise to fool GPTZero or Turnitin. The humanizer is a narrow, checkable pass against visible fingerprints—zero-width spaces, slop vocabulary, uniform sentence length—rather than snake oil. That honesty is the architecture.

Key highlights

• humanize.py and detect.py run offline with zero dependencies; no text is uploaded
• The scrubber targets 17 classes of invisible characters and 113 stock phrases stored in slop.json
• A five-check panel scores burstiness, specificity, slop density, fingerprint, and voice, weighting your weakest signal at 40% because detectors only need one tell
• /li-plan drafts a weekly calendar; /li-inbox sorts messages into lead, recruiter, peer, ask, or spam
• No API keys, no signup, no SaaS gate: skills install as plain SKILL.md files and read your custom voice.md

Caveats

• The five checks are local heuristics modeled on public detector signals, not commercial APIs; they cannot promise those verdicts
• The invisible-character pass is narrow and makes no claims about defeating cryptographic watermarking schemes
• Skip templates/voice.md and every skill defaults to the same generic tone

Verdict

A solid fit for solo operators who want structured LinkedIn help but are wary of platform bans and AI slop. Skip it if you are looking for a hands-off bot that publishes while you sleep.