skills/humanizer/SKILL.md
Remove AI writing signatures from prose. Makes text sound genuinely human-authored — varied rhythm, natural register, no AI tells. Use when drafting, editing, or reviewing any text that needs to read as if a person wrote it.
npx skillsauth add stanislavjiricek/neuroflow humanizerInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Transform AI-generated prose into text that reads as if a human wrote it. This goes beyond cutting filler — it addresses rhythm, register, word choice, and the structural patterns that mark AI output as synthetic.
Remove every word and phrase that signals machine authorship. No exceptions.
Word blacklist — replace or cut entirely:
| Banned word / phrase | Why it's a tell | |---|---| | delve, delving | No human writes "delve" unless they're writing a fantasy novel | | comprehensive, robust, nuanced | Vague intensifiers that pad without adding meaning | | crucial, pivotal, vital, essential | Overused AI emphasis words | | seamlessly, effortlessly | Fake smoothness | | leverage (as a verb) | Corporate-AI hybrid; use "use" or "apply" | | notably, importantly, significantly | AI throat-clearing before a point | | transformative, groundbreaking, revolutionary | Unjustified superlatives | | in the realm of, in the landscape of | Pompous filler | | it is worth noting, it is important to note | Cut entirely; just state the note | | this highlights, this underscores, this demonstrates | Tell the reader what it shows, don't announce that it shows it | | the [noun] landscape | Cliché geography metaphor | | moving forward, going forward | Corporate filler | | at the end of the day | Cliché | | em dash (—) | Overused by AI; replace with comma, colon, or period |
Structural blacklist:
| Pattern | Replace with |
|---|---|
| X not only A but also B | X does A and B — or split into two sentences |
| While X, Y (contrast opener) | Rewrite without "while" — state X and Y as separate facts |
| Not X — it's Y contrasts | State Y directly; cut the negation |
| Three-part lists with parallel structure | Two items or prose; three identical grammatical units sound robotic |
| Sentences starting with "It is" / "There is" / "There are" | Rewrite with a concrete subject |
| Paragraph ending with punchy one-liner summary | Vary the ending; not every paragraph needs a mic-drop |
| Wh- sentence openers ("What this means is…") | Restructure |
AI prose is metronomic. Every sentence is roughly the same length. Every paragraph has the same arc. Break it.
Rules:
Match the register to the context — but always push toward natural.
Academic writing:
Grant / proposal writing:
General prose:
The goal is not to homogenize. It is to remove machine patterns while keeping what makes the writing individual.
Run through each line:
Rate 1–10 on each dimension. Below 35/50: revise.
| Dimension | Question | |---|---| | Naturalness | Does this read like a specific person's prose? | | Rhythm | Varied, or metronomic? | | Humanity | Free of AI tells? | | Voice | Distinct, or generic? | | Density | Any word that could be cut? |
testing
This skill should be used whenever the user mentions BIDS, Brain Imaging Data Structure, BIDS conversion, BIDS validation, BIDS compliance, organizing neuroimaging data, dataset_description.json, participants.tsv, bids-validator, pybids, MNE-BIDS, or asks how to structure EEG/MEG/fMRI/iEEG/PET/DWI data for sharing or preprocessing. Also invoke when the user asks how to name scan files, what sidecar JSON fields are needed, how to set up derivatives/, or how to run fMRIPrep/MRIQC on their dataset. Invoke proactively during /data, /data-preprocess, and /data-analyze phases whenever the dataset structure is relevant to the task at hand.
tools
Phase guidance for the /meeting command. Covers meeting file structure, recurring templates, attendee resolution from profiles, Google Calendar MCP integration, agenda preparation with project context, and action-item-to-task conversion at all three levels (project, flowie, hive).
data-ai
Worker-critic agentic loop protocol — orchestrator coordinates a worker agent and a critic agent across up to 3 revision cycles to produce a vetted output for any phase.
development
Knowledge base skill — Karpathy-style LLM-maintained wiki at three levels (personal/flowie, project, team/hive). Handles ingest, query, lint, schema, and project-tagging workflows. Invoked by /flowie --wiki-* (personal), /wiki (project), /hive --wiki-* (team).